Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Research Collection
Doctoral Thesis
Mechanistic investigation of the initial phase of ozonedecomposition in drinking water and wastewaterImpact on the oxidation of emerging contaminants, disinfectionan by-products formation
Author(s): Buffle, Marc-Olivier
Publication Date: 2005
Permanent Link: https://doi.org/10.3929/ethz-a-005162223
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.
ETH Library
Diss. ETH No. 16266
Mechanistic Investigation of the
Initial Phase of Ozone Decompositionin Drinking Water and Wastewater
Impact on the Oxidation of Emerging Contaminants,
Disinfection and By-products Formation
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH
for the degree of
DOCTOR OF SCIENCES
presented by
MARC-OLIVIER BUFFLE
Dipl. Bau Ing. ETH
born May 8th 1970
citizen of Canada and Switzerland
accepted on the recommendation of
Prof. Dr. Bernhard Wehrli, examiner
Prof. Dr. Willem Koppenol, co-exammer
PD Dr. Urs von Gunten, co-exammer
Zurich 2005
Acknowledgments
Thanks to
SUEZ Environnement for financing the project, Prof Bernhard
Wehrli for complementing with an assistantship position and the Swiss taxpayer,
through the Eawag, for providing researchers with a world-class facility
Urs von Gunten for his untiring supervision throughout the PhD
process Urs shows a rare combination of unwavering availability, enthusiasm,
competency and focus, while giving his team members the freedom to forge
their own paths If there are students out there looking for a perfect supervisor,
don't look any further
the "von Gunten team" and associates Adriano Joss, Andy Peter,
Brian Sinnet, Eddi Hoehn, Gretchen Onstad, Gunyoung Park, Heinz Bader,
Jochen Schumacher, Juan Acero, Prof Juerg Hoigné, Karin Rottermann,
Laurence Meunier, Lisa Salhi, Maaike Ramseier, Manuel Polo Sanchez, Marc
Huber, Markus Boller, Max Maurer, Max Reutlinger, Michael Dodd, Olivier
Leupin, Sarahann Dow, Sébastien Meylan, Silvio Canonica, Sonja Galli, Stephan
Hug, Suzanne Metfier, Yunho Lee et al at the Eawag who made working here a
pleasure and a constant learning experience
the "CIRSEE team" of SUEZ Environnement Auguste Bruchet,
Isabelle Baudin, Jean-Michel Lamé, Marie-Laure Janex, Zdravka Do-Quang for
their scientific support and suggestions and great hospitabihty and friendship
while we visited CIRSEE, as well as Pierre-André Liechti of Ozonia
Prof Bernhard Wehrli and Prof Willem Koppenol of the ETH for
accepting to be my examiner and co-examiner, even though the subject of my
research was only remotely related to theirs, they were very supportive and had
the kindness to always be available for discussions
Trojan Technologies, and in particular present and past associates
Alan Royce, Bill Cairns, Brian Petri, Christian Williamson, Dan Gosselin, Dave
Olson, Fanborz Taghipour, Fraser McLelland, George Traubenberg, Greg
Williams, Hank Vander Laan, Harold Wright, Kuang-ping Chiu, Linda
Gowman, Linda Sealey Madjid Mohseni, Marvin DeVries, Mike Sarchese, Mike
Sasges, Phil Whiting, Pierre Sullivan, Ramm Farnood, Ron Braun, Yuri
Lawryrshyn, Ted Mao, and the sorely missed Richard Pearcey for showing me
what an industrial R&D environment should look like and giving me a strong
"taste" for water research
Prof Charles Williamson and Dr. Raghu Govardhan of the
Aerospace Department of Cornell University for demonstrating during our
collaborative investigations the true meaning of excellence in academic research
Prof Charles O'Melia of John Hopkins University, Joel Malevialle
and Jean-Michel Laine of SUEZ Environnement for putting me on the "von
Gunten" track
le très regretté Denis Mavrocordatos, pour ta constante bonne humeur
et les rires que tu nous as apporté durant ta trop courte présence parmi nous
Olivier Leupin pour quelques grands moments passés ensemble
perdus dans le brouillard sur un glacier entre deux crevasses ou coincés sur une
paroi de granite à cause d'un mauvais relais und Erich Bollinger fur die vielen
und intensiven aber Kopf luftenden Sportklettern Trainings
my family, Kalli, Tristan and Talia who have had the patience to wait
a couple of extra years before purchasing the Aston Martin (station wagon), mes
parents Reyne et Jacques, et mes grands-parents avant eux qui nous inculquèrent
la curiosité et l'apprentissage comme mode de vie et en particuher Jean-Phihbert
Buffle qui nous donna tôt le goût de la recherche aquatique et, last but not least,
Françoise, qui fut d'une aide considérable lorsque 24 heures par jour ne suffisaient
plus durant l'ultime année de rédaction
Summary
Ozone has been used in the water treatment industry for disinfection
and oxidation purposes since the late 1800s. Among oxidation
processes, typical applications are taste, odor and color removal. In
recent years, concerns following measurements of trace amounts of
contaminants in the water supplies have triggered an interest in more
specific applications such as oxidation of antibiotics, hormones,
pesticides and cyanotoxms. Current investigations show that ozone is
an efficient oxidant for many of these emerging contaminants.
When added to natural water, ozone decomposes rapidly and
secondary oxidant species, in particular HO' are formed. The
decomposition occurs in two phases, a rapid initial phase with half-
lives of the order of seconds and a second phase with half-lives of the
order of tens of minutes. The initial phase is too rapid to be resolved
with existing measurement techniques such as batch-dispenser systems.
Nevertheless, it is of considerable importance as a large fraction of
the added ozone is consumed during the first 20 seconds. A case in
point is wastewater ozonation where, under standard conditions, 100%
of the added ozone is consumed prior to 20 seconds.
The goal of this research pro|ect was to investigate and characterize
the initial phase of ozone decomposition in drinking and wastewater
and assess its impact on key ozonation processes such as the
oxidation of emerging contaminants, disinfection and formation of
by-products. To this end, an experimental system needed be designed
that could provide measurements 100 times faster than batch-
dispenser systems.
A continuous quench-flow system was developed that allows
measurements to be made 100 milliseconds after ozone addition. It
was used to characterize ozone decomposition and HO' generation
during the initial phase of ozonation. Ozone decomposition kinetics
was found be of higher order in the initial phase than during the
second phase where it generally follows an empirical first-order rate-
law Moreover, the addition of HO' scavengers did not stabilize
ozone decomposition during the initial phase. This indicates that the
initial phase is not controlled by the autocatalytic chain reaction that
is responsible for ozone decomposition during the second phase.
Hence, it suggests that the initial phase is controlled by the direct
reaction of ozone with specific moieties contained in the organic
matter. The kinetics of the initial phase was subsequently accurately
reproduced with a kinetic model that accounts for a distribution of
those reactive moieties, thereby supporting the above hypothesis.
HO' exposures measured during the initial phase were very high. In
fact, the oxidation mechanisms involved during first 20 seconds of
ozonation in natural waters and wastewaters are akin to ozone-based
advanced oxidation processes. This has important consequences for
the oxidation of micro-pollutants because compounds that are not
reactive with ozone might still go through significant transformation
due to the presence of high concentrations of HO'. Consequently,
HO'-mduced oxidation products might represent an important
fraction of all products, which is noteworthy because they might
display different degrees of biochemical rnactivation than ozone-
induced oxidation products. Ozone exposures measured in wastewater
also showed that a significant degree of disinfection can be achieved
even though ozone has entirely reacted prior to 20 seconds.
The origin of the rapid initial ozone decomposition and the high HO'
yields was investigated further. Amines and phenolic functionalities,
which are ubiquitous moieties in organic matter and react readily with
ozone when deprotonated, were shown to generate very high yield of
HO' upon ozonation. It was also found that chlormation or
brommation of secondary amines resulted in an almost complete
inhibition of HO' generation upon ozonation, while halogenation of
phenol did not.
Hence, we conclude that the initial phase of ozone decomposition is
caused by ammo compounds and activated aromatics of the organic
matter, which readily react with ozone and generate high
concentration of HO'. During the second phase, however, those
moieties have already been oxidized and ozone decomposition is
controlled by the autocatalytic radical chain reaction.
Bromate is a carcinogen that might form during the ozonation of
bromide-contammg waters. Two bromate minimization strategies
were investigated consisting of in one case a pretreatment with CIO2*
and in the other of CI2 followed by NH3 addition. Both processes are
based on the concept of decreasing HO'-mduced bromate formation
during the initial phase. When combined with a pH decrease, both
control strategies were able to decrease bromate formation roughly by
a factor of 30.
Résumé
L'ozone est utilisé depuis la fin du 19eme siècle par l'industrie du
traitement de l'eau, pour son pouvoir oxydant et désinfectant. Les buts
traditionnels de l'oxydation sont l'élimination de goût, d'odeur et de
couleur. La récente découverte de l'existence de contaminants en
concentrations traces dans les eaux de surface a accru l'intérêt pour des
applications plus spécifiques de l'ozone, telles que l'oxydation
d'antibiotiques, d'hormones, de pesticides et de cyanotoxines. Les
études actuelles montrent que l'ozonation est très efficace pour nombre
de ces nouveaux contaminants.
Lorsque l'ozone est introduit dans l'eau, il se décompose rapidement et
des espèces chimiques oxydantes secondaires se forment, en particulier
le radical hydroxyl HO'. La décomposition se produit en deux phases:
une phase initiale rapide ayant une durée de demi-réaction de l'ordre de
quelques secondes, et une seconde phase ayant une durée de demi-
réaction de l'ordre de quelques dizaines de minutes. La phase initiale est
trop rapide pour pouvoir être suivie par les techniques de mesures
classiques en réacteur discontinu. Cette phase présente pourtant un
intérêt considérable car elle consomme une grande partie de l'ozone
a|outé. L'ozonation des eaux usées en est un exemple extrême où, dans
des conditions standards, 100% de l'ozone a|outé est consommé avant
que toute mesure puisse être effectuée.
Le but de ce travail a été d'étudier et de caracténser la phase initiale de
la décomposition de l'ozone dans les eaux potables et usées et d'évaluer
son rôle sur les processus clés d'ozonation, tels que la désinfection,
l'oxydation des contaminants émergents et la formation de produits
secondaires. Pour cela, un système a dû être développé permettant
d'effectuer des mesures 100 fois plus rapides qu'en utilisant un système
en réacteur discontinu. Nous avons mis au point un système « quench-
flow en continu » qui permet d'effectuer des mesures à partir de 100
millisecondes après l'addition d'ozone. Le système a été appliqué à la
mesure de la décomposition de l'ozone et de la génération de HO' dans
les eaux de surface et les eaux usées. Il a permis de démontrer que la
décomposition de l'ozone suit une cinétique d'ordre supérieur à celle
du premier ordre généralement observée durant la seconde phase.
L'addition d'un produit consommant HO' montre que la phase initiale
n'est pas contrôlée par la chaîne de réactions auto-catalytiques qui
détermine la seconde phase. Ce résultat suggère que la phase initiale est
principalement provoquée par la réaction directe de l'ozone avec des
groupements fonctionnels spécifiques contenus dans la matière
organique. La cinétique de décomposition initiale a pu être modélisée
au moyen de distributions hypothétiques de groupements fonctionnels
réactifs, ce qui confirme cette hypothèse.
De très fortes concentrations de HO' ont pu être mesurées durant la
phase initiale, au point que dans les eaux naturelles et les eaux usées, les
processus d'oxydations correspondant aux 20 premières secondes
d'ozonation sont similaires à ceux observés lors de l'application de
procédés d'oxydation avancée. Ceci a d'importantes conséquences pour
l'élimination par l'ozone de contaminants émergents; en effet, des
composés peu réactifs avec l'ozone peuvent tout de même subir une
transformation importante grâce à la présence de fortes concentrations
de HO'. On peut donc s'attendre à ce qu'une large proportion des
produits d'oxydation soient dus à HO', ce qui pourrait avoir une
influence sur l'inactivation biochimique des molécule dont l'élimination
est recherchée. Les expositions d'ozone (JOjdt) mesurées dans les eaux
usées ont aussi permis d'estimer que le niveau de désinfection peut-être
très important même lorsque l'ozone réagi entièrement durant les 20
premières secondes.
Les causes de la rapide décomposition initiale de l'ozone et du fort
rendement de formation de HO' ont été étudiées et ont montré que les
composés aminés et phénoliques, qui sont très répandus dans la matière
organique et réagissent facilement avec l'ozone génèrent une forte
production de HO'. Nous en concluons que la phase initiale de la
décomposition de l'ozone est causée par des composés aminés et des
aromatiques activés qui réagissent directement avec l'ozone pour
former HO'. Lors de la deuxième phase, ces groupements fonctionnels
sont déjà oxydés et la décomposition de l'ozone est contrôlée par la
chaîne de réactions auto-catalytiques.
Le bromate est un cancérigène, formé durant l'ozonation d'eaux
contenant du bromure. Deux stratégies de minimisation de formation
du bromate ont été étudiées en détails. L'une d'entre elles consiste en
un prétraitement avec le dioxide de chlore, CIO2', et l'autre en une
addition de CI2, suivie de NH3. Les deux méthodes sont basées sur le
principe de la diminution de la formation de bromate par HO' durant
la phase initiale. En les combinant avec un abaissement du pH, ces
stratégies peuvent diminuer la formation de bromate par un facteur
supérieur à 30.
Table of Contents
Introduction
2 Measurement of the initial phase of O3 decomposition in 21
water and wastewater by means of a continuous
quench flow system: application to disinfection and
pharmaceutical oxidation — Water Research 2006
Ozonation and advanced oxidation of wastewater: effect 49
of O3 dose, pH, DOC and HO"-scavengers on O3 decompositionand HO" generation — Ozone Science & Engineering2006
Phenols and amines induce HO* generation during the 85
initial phase of natural water ozonation — Environmental
Science & Technology 2006
5 Enhanced bromate control during ozonation: the CI2-NH3 109
process — Environmental Science & Technology 2004
6 Enhanced bromate control during ozonation: pre- 141
oxidation with CI02* — submitted to Ozone Science & Engineering 2006
AI Moiety-Specific Oxidation of Antibacterial Molecules by 161
Aqueous Ozone: Reaction Kinetics and Application to
03-Based Wastewater Treatment — Environmental science &
Technology 2006
All Supporting Information to AI 189
1 Introduction
1.1 Background
1.1.1 Ozone in water treatment
Ozone has been used in the water treatment industry for disinfection
and oxidation purposes since the late 1800s (1-3). Among oxidation
processes, typical applications are taste, odor and color removal (4).
In recent years however, concerns following the measurements of
trace amounts of contaminants in water supplies and the natural
environment have triggered an interest in more specific applications
such as antibiotics, hormones, pesticides and cyanotoxms oxidation
during water and wastewater treatment (5-12). Current research shows
that ozone is one of the most efficient oxidants for a majority of the
emerging contaminants investigated (13-19).
1.1.2 Predicting the degree of oxidation or disinfection
In most cases, the degree of oxidation or disinfection undergone by
micro-pollutants or micro-organisms exposed to an oxidant in a
homogeneous solution can be well modeled with second-order kinetics,
i.e. the rate is first-order with respect to the compound/organism and
to the oxidant concentrations (20-23).
^H= _k* . [X] . [OJ (1)
at
where X is either a chemical compound or a micro-organism, Ox is the
oxidant concentration and k" is a second order rate constant (24).
2 Chapter 1
When integrated, eq 1 takes the form
[*h_
u. frm ^ ~m
ln(±LL)= _r .
f[O ]
. ^ or±LL
= e~* J[o,l *(2)
If [Ox] > ~ 10 x [X], and [Ox] does not self-decay significantly dunng
the reaction, so-called pseudo first-order conditions are met, i.e. [Ox] is
assumed constant, and eq 2 simplifies to
ln(i^L) = -k" [O ] • t or.Ml
= e-k" [°*] '
(3)m m«.
In eq 2 and eq 3, J[Ox]'dt or [Ox]'t are the oxidant "exposure".
é^ To insure proper disinfection during water treatment, the
regulator grants "disinfection credits" to water treatment facilities if theycan prove to be applying a certain CT value The CT concept is based on
eq 3 However, it is simplified and conservative as its practical calculation
is done by multiplying the theoretical residence time of the oxidant in a
contact chamber with the concentration of the oxidant at the outlet of
the chamber Given that most oxidants decompose during the time of
contact, a higher initial concentration of oxidant must be added to obtain
the adequate concentration at the outiet, this means that the true
"oxidant exposure" ( J [Ox] dt ) is larger than the calculated CT (25)
As will be shown below, secondary oxidants such as HO', CO3', and
O2* are generated during ozonation. Because of their very low
transient concentrations, however, only oxidants displaying very high
rate constants might compete with ozone for the oxidation of a
particular compound. Eq 4 can be used to estimate the fraction of a
compound X oxidized by one specific oxidant (e.g. HO' in eq 4).
k" .[HO*]/x(HO')
^0.[HO'] + ^3[03] + k"Q. [02-] + k"co. [C03
(4)
Introduction 3
In most cases, HO' is the only secondary oxidant that need be
considered and eq 2 becomes
ln(/^-) = -^3 -\[0,\dt-k"Ho- -\[HO-\dtY^ Jo
[X]_
-^J[03]A-Fo.J[//0*]A
[x\~e
For disinfection calculations, the effect of HO' is typically neglected.
Clearly the above equations are only valid for a perfectly mixed and
homogeneous solution. In bench scale experiments, a saturated aqueous
ozone solution is added to the water in a stirred reactor vessel. In full-
scale systems, however, ozone is usually added to the water with counter
current bubble columns or with Ventun-tube injectors. Hence, secondary
effects of, for example, imperfect mixing, gas transfer rate limitation or
particle shielded organisms must be considered and significantly
complicate the accurate modeling of full scale installation (26-35).
1.1.3 Ozone decomposition in waters containing natural
organic matter
Ozone is not stable m natural waters and wastewaters, with typical
half-lives under 60 minutes at a neutral pH. The kinetics regulating
ozone decomposition is complex. For simplification's sake, it can be
reduced into two mam phases: an initial phase with half-lives less than
twenty seconds and a second phase with a half-life between thirty
seconds and sixty minutes (Figure 1.1).
4 Chapter 1
log [03]/[03]o
Figure 1.1 The two phases of ozone decomposition in natural water
and wastewater
The semndphase, has been extensively studied (3645). The decrease of ozone
concentration over time can typically be well fitted with an empincal first-
order rate law (Figure 1.1) but the underlying mechanisms are complex. The
most widely accepted model indicates that the decomposition of ozone is
initiated by reactions with HO or HO2, which eventually generate HO'.
Hydroxyl radicals react with some moieties of the organic matter to generate
superoxide, O2* • Superoxide reacts specifically with ozone to generate the
ozonide radical, O3', which decays instantaneously to HO' (Figure 1.2), and
so on (41,44). The overall mechanism has been called "autocatalytic ozone
decomposition" or "radical-type chain reaction". Alkalinity, pH, temperature,
type and concentration of the dissolved organic matter are crucial parameters
influencing the rate of the chain reaction (Figure 1.2) (4447).
As seen above, dissolved organic matter (DOM) plays a central role in the
second phase of ozone decomposition. During the autocatalytic decay it
can act as an initiator, promoter or inhibitor (Figure 1.2) (41,44). A large
number of studies (46-52) have been published trying to link vanous DOM
chemical or physical charactenstics to ozone decay. However, due to the
complexity of DOM composition (53-56) and ozone reaction pathways, it
is impossible to reach an accurate deterministic descnption of the
decomposition of ozone in natural waters.
Introduction 5
03
4-[m+]*<--
v kh+ra;,
2 3 10 «Mis'
c
'03-H+ J5 10«M
T. 1 & «n
16 109M's' IB00"
R*
T *S&~
Figure 1.2 Mechanisms involved in the decomposition of ozone in
DOM-containing water, adapted from (44)
The initialphase has received much less attention, even though it is
crucial from a system efficiency standpoint as well as to the
understanding of oxidation mechanisms during ozonation. The
difficulty m measuring ozone concentration in such short time
frames had significantly hindered studies until now.
Due to its rapidity, the initial phase has been called "instantaneous
ozone demand" (IOD) (51), "instantaneous ozone consumption"
(57,58), "initial rapid ozone consumption" (46). In (57,58), Hoigné
and Bader teach that studies on ozonation of natural waters should
always contain two standard measurements to allow fair
comparisons to be made: the ID and the second half-life of ozone,
where ID is defined as the amount of ozone consumed during the
6 Chapter 1
first 20 seconds of ozonation. The second half-life characterizes the
autocatalytic decay phase of the process.
é^ The duration of 20 seconds is operationally denned It correspondsto the first possible ozone concentration measurement in standard ozone
kinetics experiments using a batch reactor vessel with dispenser
Westerhoff et al. (50) investigated the initial ozone demand (which they
call Aoi) as a separate variable from the second phase rate constant k03-
Aoi's sensitivity to the vanous DOM-isolates was similar to that of ko3-
However, one noteworthy difference between ko3 and Aoi was the
sensitivity of these parameters to the presence of a HO'-scavenger.
Scavenging HO' had significantly more impact on ko3 than on Aoi-
Westerhoff et al. (50) did not characterize the actual initial ozone
kinetics nor did they propose an initial mechanism.
Park et al. (51) also describe the importance of understanding the initial
phase of ozonation. An apparatus was especially developed to
investigate the initial reaction, however its time-resolution is not tested
or calibrated and the authors stop short of actually characterizing the
initial kinetics, merely publishing a table comparing initial demand to
total organic carbon concentrations.
Elovitz and von Gunten (46,47) give Ret values (Re, = J[HO']dt/J[03]dt,i.e. hydroxyl radical exposure to ozone exposure) for the initial and the
second phase of ozone decomposition. Results show significantly
higher relative concentrations of HO' during the initial phase.
Although an extensive literature search was conducted, published data
charactenzing the kinetics of ozone decomposition during the initial
phase could not be found.
Introduction 7
1.1.4 Ozone concentration measurement
In normal water treatment applications and at standard conditions,
ozone concentration cannot be assumed to remain constant during
the entire duration of the ozonation process. Hence, to predict the
degree of oxidation or disinfection following ozonation in specific
waters, eq 2 (or eq 5) should be used instead of eq 3. To obtain
ozone exposure (J[Os]'dt in eqs 2, 5) O3 concentration must be
measured and integrated over time.
Typically, bench scale measurements of ozone concentration in
water are performed with batch systems (57,58). The natural water is
stirred while aqueous ozone is added and using a dispenser, water
samples are injected into Indigo-contammg vials at regular intervals.
It takes roughly 20 seconds for the first sample to be taken, hence
the name: "instantaneous ozone demand". The ozone concentration
is calculated based on the decolounzation of mdigo which turns
transparent upon its reaction with ozone in a one to one
stoichiometry at an acidic pH (59). Figure 1.3 shows typical results
obtained when measuring ozone concentration in water (a) and
wastewater (b) with a batch-dispenser system. While there might be
a significant instantaneous demand in natural water, the second
phase is well resolved and J[Os]'dt (area under the concentration
curve) can be calculated. In wastewater, however, the demand is so
large that ozone concentration cannot be measured and no
prediction can be made.
Figure 1.3 clearly shows the need for a faster method to be
developed if the initial phase in natural water and ozone
decomposition in wastewater are to be measured.
Chapter 1
[O3]
O3 dose ^>.
1st batch meas - U
(a) Ozone in Natural Water
ylOD Instantaneous Ozone Demand
2nd phase
20sec 10-100min
[O3] (b) Ozone in Wastewater
in wastewater IOD = ~03 dose
^N—t-20sec
-v' time
10-100min
Figure 1.3 Ozone measured with a batch-dispenser system in (a)natural and (b) wastewater
Commercial quench flow systems allow ozone measurements after
~1 millisecond (Bio-Logic, Applied-Photophysics, Olis, KinTek (60-63)),
however these systems are based on single-push syringes which
significantly limits sampling volumes for post-sampling analysis. It is
therefore important to develop a system that allows the sampling of
large enough volumes for subsequent SEC, HPLC, GC or IC analyses.
Introduction 9
1.2 Research Objectives
The ob|ectives of this research pro|ect can be subdivided in three categones.
1.2.1 Method development objectives
a. Development of a method/apparatus for the simultaneous
measurement of rapid ozone and hydroxyl radical kinetics
(<30s) in waters containing DOM-loadmgs ranging from natural
waters to secondary wastewater effluent.
b. A "static" method and a guideline for the practical determination
of the initial O3 exposure in drinking and wastewater.
1.2.2 Scientific objectives
c. Characterization and mechanistic description of the initial ozone
decay kinetics in various DOM-contatning waters.
d. Charactenzation of the relative exposures of HO' and ozone (R«)
in various DOM-contatning waters during the initial decay phase.
e. A mechanistic explanation for the high relative concentrations
of HO' created during the initial ozone decay phase in DOM-
contammg waters.
1.2.3 Engineering objectives
f A tool for modeling initial ozone decay kinetics.
g. A control option to limit bromate formation using mechanistic
knowledge of the initial ozone decay phase.
h. Measurement and modeling of the degree of oxidation of
pharmaceuticals (e.g. antibiotics, hormones) during the initial
ozone decay phase for drinking and wastewater.
1. Measurement and modeling of the degree of microbial inactivation
during the initial ozone decay phase for dnnktng and wastewater.
10 Chapter 1
1.3 Thesis Layout
The present thesis is based on articles published, submitted or
expenmental work performed pnor to the PhD defense examination.
Each article/chapter is its own entity; it is therefore important to
explain how they relate to one another.
Chapter 1 —Introduction— describes the background of this pro|ect,
enumerates the research ob|ectives, explains the connections between
the chapters, summarizes results and finally presents a research outlook.
Chapter 2 —Measurement of the Initial Phase of O^one Decomposition in Water
and Wastewater by Means of a Continuous Quench Flow System: Application to
Disinfection and Pharmaceutical Oxidation (Wat. Res., 2006 (64))— descnbes
the development of the Continuous Quench Flow System to measure
the initial phase of ozone decomposition in drinking water and
wastewater. It then shows applications of the measured oxidant
exposures to predict the oxidation of pharmaceutical compounds and
the mactivation of microorganisms in wastewater.
Chapter 3 —Ozonation and Advanced Oxidation of Wastewater: Effect of 03
Dose, pH, DOC and HO'-scavengers on O^one Decomposition and HO'
Generation (O^one Sa. Eng., 2006 (65))— is a parametric investigation of
ozone decomposition and HO' generation m the same wastewaters as
in Chapter 2. Chapter 3 also includes some attempts to mechanistically
model the initial phase.
Chapter 4 —Phenol and Amine-lnduced HO' Generation During the Initial
Phase of Natural Water Ozonation (Environ. Sa. Technol, 2006 (66))— gives
a mechanistic explanation for the high HO' yield and high rate of
ozone decomposition measured dunng the initial phase in Chapter 2
and 3. Chapter 4 also investigates the effect of pre-chlonnation and
pre-bromrnation on the generation of HO' upon ozonation.
Introduction 11
Chapter 5 —Enhanced Bromate Control During Ozonation: The Chlorine-
Ammonia Process (Environ. Sa. Technol, 2004 (67))— investigates and
characterizes the mechanisms involved in a new control strategy to
minimize bromate formation. The mechanisms can be well explained
using the base of knowledge acquired in the preceding chapters.
Chapter 6 —Enhanced Bromate Control during Ozonation: Pre-oxidation with
CIO2 (submitted to Ozpne Sa. Eng., 2006 (68))— descnbes another method
for bromate minimization based on the pre-oxidation of the water
matrix by CIO2'. An important part of the mechanisms can be explained
using the base of knowledge acquired in the preceding chapters.
Appendix —Moiety-Specific Oxidation of Antibacterial Molecules by Aqueous
Ozpne: Reaction Kinetics and Application to Ozpne-Based Wastewater Treatment
(Environ. Sa. Technol, 2006 (19))— shows some interesting applications
of the knowledge acquired in the preceding chapters to understand the
oxidation of antimicrobial agents in wastewater.
1.4 Results Summary
1.4.1 Methods development
a. A continuous quench-flow system was developed, which can
start measurements 100 milliseconds after ozone addition. Rate
constants measured with this system were within a few % of
published values. The system was successfully applied to
measure ozone decomposition and HO' generation in surface
waters and wastewaters.
b. For full scale system, the use of an ozone probe compound was
suggested that would work similarly to HO'-probe, pCBA.
Huber et al (15) tested the concept during wastewater pilot
12 Chapter 1
experiments but the prediction was not accurate for a number
of compounds. Further theoretical investigations on the effect
of imperfect reactor mixing must be undertaken prior to
endorsing this method. At the bench scale, however, as long as
sufficient mixing is guaranteed, the use of an ozone probe
compound is straight forward.
1.4.2 Scientific results
c. Although ozone decomposition follows apparent first order
kinetics during the second phase, its apparent first-order rate
constant increases with a power function when approaching t=0,
both m wastewaters and in surface waters. The addition of HO'
scavengers demonstrated the initial phase not to be controlled
by the autocatalytic chain reaction, responsible for ozone
decomposition during the second phase.
d. Very high HO' exposures could be measured during the initial
phase, to the point that the first 30 seconds of ozonation in natural
waters and wastewaters can be described as an advanced oxidation
process — AOR R« (=J[HO']dt/J[03]dt) follows a power function
when approaching t = 0. In fact, the transient HO' concentrations
are 100 times larger than in lab-scale UV-H2O2 systems.
e. The causes for the rapid initial ozone decomposition and high
HO' yields were investigated further. Amines and phenolic
compounds, which are ubiquitous m the NOM and react readily
with ozone when deprotonated, were shown to generate very
high yields of HO' upon ozonation. Chlonnation or
bromrnation of secondary amines almost completely hindered
HO' generation upon ozonation, while halogenation of
phenolic compounds did not.
Introduction 13
1.4.3 Engineering results
f. Initial ozone kinetics could be well fitted with a kinetic model
using distributions of NOM moieties. Using the fitted
distribution, changes in ozone dose could be well predicted by
the model. This confirms that the initial phase is mostly due to
direct reactions with specific functional groups contained in the
organic matter.
g Two bromate minimization strategies consisting of in one case a
pretreatment with CIO2* and in the other, CI2 followed by NH3
addition, were investigated and further developed. Both
methods are based on the principle of blocking bromate formed
by HO' dunng the initial phase. When combined with a lowering
of the pH, both control strategies can decrease bromate
formation by a factor larger than 30.
h. Water and wastewater were spiked with the antiepileptic drug
carbamazepme. Its degree of oxidation was measured and
compared to predictions based on measured ozone and hydroxyl
radical exposures and on published rate constants (with eq 5).
The model was able to accurately predict the measured results. It
was also clearly demonstrated that for accurate predictions, HO'
need be taken into account.
1. Based on ozone exposure measurements in wastewaters,
modeling of inactivation of vanous microorganisms indicated
that many microorganisms can be inactivated to a significant
degree, even if no ozone residual is left 20 seconds after ozone
addition. An exception is Cryptosporidium parvum oocysts which
require significantly higher ozone exposure to be inactivated.
14 Chapter 1
1.4.4 Implications for the water treatment industry
There are a number of consequences for the water treatment industry
that can be denved from this research pro|ect.
Ozone is an advanced oxidation process (AOP) in wastewater and
during the initial phase in drinking water, i.e. HO' plays a very
important role in the oxidation of compounds that are not extremely
reactive with ozone. This finding is two fold, on one hand it might be
positive because HO' will oxidize ozone-refractory compounds, but on
the other hand, it might have the disadvantage of generating more
unknown byproducts. For example better knowledge of the
biochemical activity of metabolites generated by HO' upon ozonation
of pharmaceuticals in the water must now be gained.
The exposure to ozone in wastewater, even when no ozone residual is
measurable after 20 seconds, can be considerable from a disinfection
standpoint. North Amencan wastewater disinfection requirements could
be met easily with very low ozone doses. Given the fact that ozone
readily de-activates a large number of estrogenic compounds and that
the discharge of hormones from secondary effluent into the
environment might be linked with significant environmental damage, the
use of ozone as a final step in wastewater treatment might be beneficial.
Two bromate control strategies were developed that can essentially
reduce bromate concentrations below the existing detection limit.
While resolving the issue of bromate formation, both techniques
require the addition of another oxidant before ozonation which, in
turn, will automatically generate oxidation by-products. These by¬
products (such as THMs in the CI2-NH3 process) were shown to be
well below the dnnkmg water standard but from a public health
standpoint one might question if such "chemical acrobatics" are truly
a benefit for the consumer.
Introduction 15
1.5 Research Outlook
As mentioned above, investigation of the biochemical activity removal
by HO' oxidation of pharmaceutical compounds is important if a case
is to be made for the use of ozone in wastewater to remove estrogenic
and antibactenal molecules.
Dunng this investigation the absorption of oxidized water at 285 nm in
wastewater could be directly correlated with the ozone exposure. Such
measurements should be made on a large number of wastewaters,
because if confirmed it would represent a very simple method for a
utility to obtain ozone exposure. Also, normalizing with the
concentration value of DOC seemed to have a unifying effect across
vanous waters for some cntical ozonation parameters, this should be
further investigated.
The development of an easily-analyzable ozone probe, or probe senes
(when the ozone exposure cannot be guessed) and a lab-scale system
allowing rapid dosage and mixing of ozone into a wastewater
containing the ozone probe could give engineers the ability to
determine ozone exposure off-line.
The use of such ozone probe in large scale system is promising,
however, limitations due to imperfect mixing should be investigated.
The development of a quench-flow system that would enable a
continuous measurement of the decrease of ozone concentration,
would potentially allow the observation of kinetic steps induced by
individual reactive moieties (see Chapter 3 for discussion).
In this investigation, all experiments were done in homogeneous one-
phase flows, hence, potential difficulties linked to mass transfer
limitations could be neglected. Given the rapidity of the initial ozone
16 Chapter 1
reactions, mass transfer simulation should be performed to investigate
potential chemical reaction limitation in wastewater (e.g. with phenolic
compounds). The initial phase lasts only seconds. Thus, most of it
happens in the bubble column, and mixing is likely to play an important
role beside mass transfer issues. CFD should be used to investigate
those effects.
A number of molecules may generate low yields of HO' upon
ozonation even though the main mechanism may not. Dunng this
investigation, an increase in HO' generated was measured when mtnte
containing water was ozonated (data not reported) even though the
main mechanism is known to be an oxygen atom transfer. Such effects
should be further investigated.
1.6 References
1 Marinier, Abraham Stérilisation des eaux par l'ozone, Société Industrielle de
l'Ozone Paris, 1900
2 Imbeaux, E Qualités de l'eau et moyens de correction, Dunod Paris, 1935
3 Buffle, J -Ph ,La désinfection des eaux destinées à la consommation
Bulletin Soc. Lyon. Eaux 1977, 49, 21-32
4 Langlais, B, Reckhow, D A, Brink, D R O^one m water treatment:
application and engineering, Lewis Publisher, Inc,1991
5 Clara, M , Strenn, B, Kreuzinger, N, Carbamazepme as a possible
anthropogenic marker in the aquatic environment investigations on the
behavior of Carbamazepme in wastewater treatment and during
groundwater infiltration Water Research 2004, 38, 947-954
6 Cleuvers, M, Aquatic ecotoxicity of pharmaceuticals including the
assessment of combination effects Toxicology Letters 2003, 142, 185-194
7 Daughton, C G, Ternes, T A, Pharmaceuticals and personal care
products in the environment agents of subtie change^ Environ. Health
Perspect. 1999, 107, 907-937
8 Ferrari, B, Paxeus, N, Lo Giudice, R, Pollio, A, Game, J,
Ecotoxicological impact of pharmaceuticals found in treated wastewaters
study of carbamazepme, clofibric acid, and diclofenac Ecotox. Environ.
Safe. 2003, 55, 359-370
Introduction 17
9 Jos, A , Repetto, G, Rios, J C, Hazen, M J, Molero, M L
,del Peso, A ,
Salguero, M, Fernandez-Freire, P, Perez-Martin, J M, Camean, A,
Ecotoxicological evaluation of carbamazepine using six différent model
systems with eighteen endpoints Toxicology m Vitro 2003, 17, 525-532
10 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 Environ. Sa. Technol. 2002, 36, 1202-1211
11 LavtUe, N, Ait-Aissa, S, Gomez, E, Casellas, C, Porcher, J M ,
Effects
of human pharmaceuticals on cytotoxicity, EROD activity and ROS
production in fish hepatocytes Toxicology 2004, 196, 41-55
12 Purdom, C E, Hardiman, P A, Bye, V J, Eno, N C, Tyler, C R,
Sumpter, J P, Estrogenic effects of effluents from sewage treatment
works Chem. Ecol. 1994, 8, 275-285
13 Westerhoff, P, Yoon, Y, Snyder, S, Wert, E, Fate of Endocrine-
Disruptor, Pharmaceutical, and Personal Care Product Chemicals duringSimulated Drinking Water Treatment Processes Environ. Sa. Technol. 2005
14 Ternes, T A, Stuber, J, Herrmann, N, McDowell, D, Ried, A,
Kampmann, M, Teiser, B, Ozonation a tool for removal of
pharmaceuticals, contrast media and musk fragrance from wastewater^
Wat. Res. 2003, 37, 1976-1982
15 Huber, M M, Goebel, A , Joss, A , Hermann, N , Loeffler, D, McArdell,
C S, Ried, A , Siegrist, H , Ternes, T A
,Von Gunten, U, Oxidation of
Pharmaceuticals during Ozonation of Municipal Wastewater Effluents A
Pilot Study Environmental Science and Technology 2005, 39, 4290-4299
16 Huber, M M, Korhonen, S
, Ternes, T A,von Gunten, U, Oxidation of
pharmaceuticals during water treatment with chlorine dioxide Water
Research 2005, 39, 3607-3617
17 Huber, M M, Canonica, S, Park, G-Y, von Gunten, U, Oxidation of
pharmaceuticals during ozonation and advanced oxidation processes
Environ. Sa. Technol. 2003, 37, 1016-1024
18 Onstad, G D, Strauch, S, Menluoto, J, Codd, G, von Gunten, U,
Selective Oxidation of Cyanotoxins by Ozonation Treatment Environ. Sa.
Technol. submitted
19 Dodd, M C, Buffle, M-O, von Gunten, U, Moiety-specific oxidation of
antibacterial molecules by aqueous ozone Reaction kinetics and relevance
to ozone-based wastewater treatment Environ. Sa. Technol. in press, 2006
20 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water - I Non-dissociating organic
compounds Wat. Res. 1983, 17, 173-183
18 Chapter 1
21 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water - II Dissociating organic compoundsWat. Res. 1983, 17, 185-194
22 Hoigné, J, Bader, H, Haag, W R, Staehelin, J, Rate constants of
reactions of ozone with organic and inorganic compounds in water - III
Inorganic compounds and radicals Wat. Res. 1985, 19, 993-1004
23 von Gunten, U, Ozonation of drinking water Part II Disinfection and
by-product formation in presence of bromide, iodide and chlorine Wat.
Res. 2003, 37, 1469-1487
24 Levenspiel, O Chemical Reaction Engineering, 3rd Edition, 1998
25 USEPA Disinfection, Profiling and Benchmarking Guidance Manual., 1999
26 Do-Quang, Z, Cockx, A, Laine, J-M Use of CFD modeling &
simulation tools for the design of different ozone contacting systems
27 Do-Quang, Z, Laine, J -M , Duguet, J -P, Roustan, M Latest advances in
the development of new simulation tools for the design and operation
control of ozone reactors Kyoto, Japan
28 Heyouni, A, Roustan, M, Do-Quang, Z, Hydrodynamics and mass
transfer in gas-liquid flow through static misers ChemicalEngineering Science
2002, 57, 3325-3333
29 Janex, M-L, Savoye, P, Roustan, M, Do-Quang, Z, Laine, J-M,
Lazarova, V, Wastewater disinfection by ozone influence of water qualityand kinetics modeling O^one Sei. Eng. 2000, 22, 113-121
30 Roustan, M, Debellefontaine, H, Do-Quang, Z, Duguet, J -P,
Development of a method for the determination of ozone demand of a
water O^one: Science & Engineering 1998, 20, 513-520
31 Roustan, M , MalLevialle, J, Roques, H , Jones, J P, Mass transfer of ozone to
water a fundamental study O^one: Science e^Engineering 1980, 2, 337-344
32 Gurol, M D, Singer, P C, Dynamics of the ozonation of phenol II
Mathematical simulation Water Research 1983, 17, 1173-1181
33 Singer, P C, Gurol, M D, Dynamics of the ozonation of phenol I
Experimental observations Water Research 1983, 17, 1163-1171
34 Rakness, K L, Corsaro, K M, Hale, G, Blank, B D, Watewater
disinfection with ozone - Process control and operating results O^one Sei.
Eng. 1993, 15, 497-514
35 Paraskeva, P, Graham, N J D, Ozonation of municipal wastewater
effluents Wat. Envir. Res. 2002, 74, 569-581
36 Buehler, R E, Staehelin, J, Hoigné, J, Ozone decomposition in water
studied by pulse radiolysis 1 H02/02- and H03/03- as intermediates
/. Phys. Chem. 1984, 88, 2560-2564
Introduction 19
37 Lesko, T M , Colussi, A J, Hoffmann, M R, Hydrogen Isotope Effects
and Mechanism of Aqueous Ozone and Peroxone Decompositions
Journal of the American Chemical Soaety 2004, 126, 4432-4436
38 Sehested, K, Corfitzen, H , Holcman, J, Fischer, C H, Hart, E J, The
primary reaction in the decomposition of ozone in acidic aqueous
solutions Environ. Sa. Technol. 1991, 25, 1589-1596
39 Staehekn, J, Hoigné, J, Decomposition of ozone in water Rate of
initiation by hydroxyde ions and hydrogen peroxide Environ. Sa. Technol.
1982, 16, 676-681
40 Staehelin, J, Buehler, R E, Hoigné, J, Ozone decomposition in water
studied by pulse radiolysis 2 OH and H04 as chain intermediates J. Phys.Chem. 1984, 88, 5999-6004
41 Staehelin, J , Hoigné, J , Decomposition of ozone in water in the presence
of organic solutes acting as promoters and inhibitors of radical chain
reactions Environ. Sa. Technol. 1985, 19, 1206-1213
42 Westerhoff, P, Song, R, Amy, G, Minear, R, Applications of ozone
decomposition models O^one: Science & Engineering 1997, 19, 55-73
43 Chelkowska, K, Grasso, D, Fabian, I, Gordon, G, Numerical simulations of
aqueous ozone decomposition O^one Sa. Eng. 1992, 14, 33-49
44 Hoigné, J In The Handbook of Environmental Chemistry, Hrubec, J, Ed,
Sponger Verlag, 1998, Vol 5, pp 83-141
45 von Gunten, U, Ozonation of drinking water Part I Oxidation kinetics
and product formation Wat. Res. 2003, 37, 1443-1467
46 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios duringozonation processes I The Ret concept O^one Sa. Eng. 1999, 21, 239-260
47 Elovitz, M S, von Gunten, U, Kaiser, H-P, Hydroxyl radical/ozone
ratlos during ozonation processes II The effect of temperature, pH,
alkalinity and DOM properties O^one Sa. Eng. 2000, 22, 123-150
48 Kato, Y, Monoka, T, Hoshikawa, H , Okada, M , Moniwa, T In Proceeding
of the 13th o^one world congress, October 26th-34th, Kyoto, Japan Kyoto, Japan,
1997, Vol l,pp 387-391
49 Bezbarua, B K, Reckhow, D A In Proceeding of the 13th o^one world congress,
October 26th-34th, Kyoto, Japan Kyoto, Japan, 1997, Vol 1, pp 337-342
50 Westerhoff, P, Aiken, G, Amy, G, Debroux, J, Relationship between the
structure of natural organic matter and its reactivity towards molecular
ozone and hydroxyl radicals Wat. Res. 1999, 33, 2265-2276
51 Park, H-S, Hwang, T-M, Kang, J-W, Choi, H, Oh, H-J,Characterization of raw water for the ozone application measuring ozone
consumption rate Wat. Res. 2001, 35, 2607-2614
52 Ho, L, Newcombe, G, Croué, J-P, Influence of the character of NOM
on the ozonation of MIB and geosmin Wat. Res. 2002, 36, 511-518
20 Chapter 1
53 Buffle, J Complexation reactions m aquatic systems: an analytical approach, Ellis
Horwood Limited Chichester, 1988
54 Buffle, J, Huang, P M, Senesi, N Structure and surface reactions of soil
particles, John Wiley & Sons, 1998, Vol 4
55 Frimmel, F H, Abbt-Braun, G, Heumann, K G, Hock, B
, Luedemann,
H D, Editors Refractory Organic Substances m the Environment, 2002
56 Dignac, M-F, Caractérisation chimique de la matière organique au cours
du traitement des eaux usées par boues activées Thèse de Doctorat de
l'UniversitéPans L71998
57 Hoigné, J, Characterization of water quality entern for ozonation processes
Parti Minimal set of analytical data O^one Sa. Eng. 1994, 16, 113-120
58 Hoigné, J, Bader, H, Characterization of water quality criteria for
ozonation processes Part2 Lifetime of added ozone O^one Sa. Eng. 1994,
16, 121-134
59 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Wat. Res. 1981, 15, 449-456
60 Apphed_Photophysics_Ltd 203/205 Kingston Road, Leatherhead SurreyKT22 7PB, United Kingdom
61 Bio-Logic_ScienceInstrumentsSA l,rue de l'Europe, F-38640 - CLAIX - France
62 KinTekCorporarion 7604 Sandia Loop, Suite C, Austin, TX 78735, USA
63 Ohs_Inc, 130 Conway Dnve, Suites A & B, Bogart, GA, 30622 USA
64 Buffle, M-O, Schumacher, J, Salhi, E, Jekel, M, von Gunten, U,
Measurement of the Initial Phase of Ozone Decomposition in Water and
Wastewater by Means of a Continuous Quench Flow System Applicationto Disinfection and Pharmaceutical Oxidation Wat. Res. accepted, 2006
65 Buffle, M-O, Schumacher, J, Meylan, S, Jekel, M, von Gunten, U,
Ozonation and Advanced Oxidation of Wastewater Effect of 03 Dose,
pH, DOM and HO"-scavengers on Ozone Decomposition and HO"
Generation O^one Sa. Eng. accepted, 2006
66 Buffle, M-O, von Gunten, U, Phenols and Amines Induce HO"
Generation During the Initial Phase of Natural Water Ozonation Environ.
Sa. Technol. accepted, 2006
67 Buffle, M-O, Galli, S, von Gunten, U, Enhanced Bromate Control
during Ozonation The Chlorine-Ammoma Process Environ. Sa. Technol.
2004, 38, 5187-5195
68 Buffle, M-O, Galli, S, von Gunten, U, Enhanced Bromate Control
during Ozonation Pre-oxidation with C102" O^one Sa. Eng. submitted
2 Measurement of the Initial Phase of Ozone
Decomposition in Water and Wastewater byMeans of a Continuous Quench Flow System:
Application to Disinfection and Pharmaceutical Oxidation
Marc-Olivier Buffle, Jochen Schumacher,
Elisabeth Salin, Martin Jekel, Urs von Gunten,
Water Research, 2006
2.1 Abstract
Due to a lack of adequate experimental techniques, the kinetics of the first
20 seconds of ozpne decomposition in natural water and wastewater is still poorly
understood. Introduang a Continuous Quench Flow System (CQTS), measurements
starting 350 milliseconds after ozpne addition are presentedfor the first time. Very
high HO' to 0, exposures ratios (R^, — }HO'dt/\03dt) reveal that the first
20 seconds of ozonation present oxidation conditions that are similar to ozpne-based
Advanced Oxidation Processes (AOP). The oxidation of carbamazepme can be
accurately modeled using 03 and HO' exposures measured with CQfS during
wastewater ozonation. These results demonstrate the applicability of bench scale
determined second-order rate constantsfor wastewater ozonation. Important degrees of
pharmaceutical oxidation and microbial inactivation are preàcted, indicating that a
significant oxidation potential is available during wastewater ozonation, even when
ozpne is entirely decomposed in thefirst 20 seconds.
2.2 Introduction
Recent studies have shown that many pharmaceutical compounds can
be detected in the effluent of wastewater treatment plants (1,2).
Concurrently, an increasing body of evidence indicates that antibiotics,
hormones and antiepileptics are responsible for microbial resistance
building, féminisation of higher organisms and ecotoxicological issues
in the aquatic environment, respectively (3-8). Renewed interest m
ozonation was spurred after it was recently discovered that the
22 Chapter 2
oxidation reactions of ozone with many pharmaceuticals exhibit very
large second-order rate constants (9). Moreover, it appears that the
moieties of pharmaceutical molecules that are the most easily attacked
by ozone often are keys to the molecules' biochemical-activities (10).
For example, although ozonation of 17a-ethinylestradiol does not lead
to full mineralization of the compound at practical doses, it does
effectively remove the compound's estrogenicity by specifically
targeting receptor active moieties, thereby generating innocuous
oxidation products (11). Subsequent pilot scale experiments performed
at a wastewater treatment plant confirmed that fast reacting
pharmaceuticals were indeed degraded almost entirely at very cost
effective doses (i.e. > 2 mg03/L) (12).
A significant obstacle m the use of kinetic models for oxidation
performance predictions m wastewater is the difficulty associated
with the measurement of oxidant exposures. High concentrations of
organic matter, certain moieties of which react very rapidly with
ozone, prevent the use of standard analytical techniques. This has
been the mam hindrance to investigations of ozone decomposition
kinetics m wastewater.
The standard experimental protocol for the characterization of
drinking water ozonation recommends the use of a batch reactor
system (13,14). Aqueous ozone is added to the water, stirred and water
is sampled at regular intervals starting roughly 20 seconds after ozone
addition. The amount of ozone consumed before the first
measurement (at ~20 seconds) is defined as Instantaneous Ozone
Demand (IOD) and is represented by a straight vertical line m
concentration versus time plots (see Figure 2.1a).
Continuous quench flow system 23
[03 a) Natural Water
20 sec
[o3
I IOD
• tew
|03dt
mwm*....mm.........
I
20 min
b) Wastewater
I IOD
20 sec 20 min
Figure 2.1. Ozone decomposition when observed with a batch
system (a) in natural water (b) in wastewater, ozone "disappears"
entirely prior to the first measurement
Integration of the ozone concentration over time gives the ozpne
exposure (JOjdt m eq 1, shaded area m Figure 2.1a) from which the
oxidation of chemical substances (P m eq 2) or the inactivation of
micro-organisms (IV in eq 3) can be calculated given known second-
order rate constants. HO' exposure (jHO'dt) is typically back-
calculated from the degradation of the HO'-probe, pCBA (15).
24 Chapter 2
Ct .= [C .-dt [M-s] (1)
0
Ct.
' oxidant exposure (shaded area m Figure 2.1a) [M-s]03 ,HO
C.
= fit) ' oxidant concentration (dotted line in Figure 2.1a) [M]
tR: reaction time (1200 s m Figure 2.1a) [s]
[P] l[P\ = e~khi Ct0i ~k"HO- CtHO-[ - ] (2)
[N]/[N]0=e-k'°>ct°> [-] (3)
k".
= second-order rate constant [M h *]
When applying cost effective ozone doses to wastewater, however, ozone
is entirely consumed prior to 20 seconds (IOD > ozone dose) and JOjdt
cannot be calculated (Figure 2.1b). Up to now IOD has therefore been
considered "wasted" ozone, an inherent inefficiency of ozonation
systems. One could assume a linear decrease between [Oj]o and
[O3]20s = 0, but this calculation severely overestimates the actual ozone
exposure. Conversely, assuming JOjdt = 0 is overly conservative as it
predicts no oxidation or disinfection.
Consequently engineers have used empirical techniques to descnbe
wastewater ozonation (16,17). Often, experiments are run on pilot scale
reactors and results relative to the degradation of certain compounds or
inactivation of particular micro-organisms are difficult, if not impossible to
extrapolate to other conditions, compounds or micro-organisms. Huber et al.
(12) tried to circumvent this difficulty by using the extent of a compound's
degradation to back-calculate ozone exposure. The extracted exposure was
then used to model the oxidation of other compounds. However, significant
differences between predictions and expérimental results were observed.
Continuous quench flow system 25
In this paper we introduce a continuous quench flow system (CQFS), a
bench-scale experimental technique developed to measure the first
20 seconds of ozone decomposition in natural water and in wastewater.
Results showing ozone decomposition and HO' generated during the
first 20 seconds of surface water and wastewater ozonation are
presented for the first time. The degree of oxidation of carbamazepme
(an antiepileptic) in wastewater is compared to predictions based on
measured JOjdt and jHO'dt. Finally, the measured oxidant exposures
are used to predict the degree of oxidation and inactivation of various
pharmaceuticals and micro-organisms, respectively.
2.3 Materials and Methods
2.3.1 Reagents
All reagents were of analytical grade. All solutions were prepared
with MilliQ water with a resistivity> 18 MQ-cm. Aqueous ozone was
prepared as described elsewhere (18); stock solution concentration
was typically 1.6 mM.
2.3.2 Water characteristics
Waters were buffered with borate for all experiments at pH 8 and
phosphate for lower pH and ad|usted with NaOH or H2SO4. pH was
controlled at the beginning and end of each expenment and was withm
+0.05 pH unit. All experiments were performed at 22 + 1°C. All waters
(see Table 2.1) were filtered at 0.45 urn (cellulose nitrate filters) and
kept at 5°C for the entire duration of the investigation. The Opfikon
wastewater treatment plant (Zurich, Switzerland) is descnbed elsewhere
(19); the water was obtained post sand filtration. Berkn wastewater was
obtained from the effluent of a secondary treatment tram at the
Ruhleben WWTP, Germany (20). Berkn drinking water was obtained at
26 Chapter 2
a household tap and Lake Zurich water was collected from the raw
water intake of the Zurich drinking water treatment plant, 30 meters
below the lake's surface (18).
Table 2 1 Water quality parameters of the tested -waters
DOC NO3 NH3/NH4+ NO2
Water mgC/La2M"m
mgN/L MgN/L MgN/L
type (mMC)m
(jjJM) (jjJM) (jiM)
OpfikonWW
45
(375)12 7
28
(2000)330 (24)
62
(4 4)
Berlin
WW
85
(708)23 9
57
(407)52 (4)
65
(4 6)
Zurich
LW
14
(117)31
0 77
(55)5 (0 36)
1 1
(0 08)
Berlin 489
04nd nd
DW (333) (29)
Alkalinity pH
(mM) ()
36 79
34 80
24 78
39 79
WW -waste-water effluent, DW drinking -water, LW lake -water
2.3.3 Analyses
Phenol was measured using HPLC with fluorescence detection (11).
Carbamazepine was measured with HPLC with UV detection (9). HO'
exposure was back-calculated using the oxidation of para-chlorobenzoic
acid (pCBA), analyzed with HPLC (15). Ozone was measured online with
a Vanan Cary 100, either directly at 258 nm (e = 3000 M lcca l) or with
the indigo reagent at 600 nm (e = 20'000 M 1cm a) (21).
2.3.4 Continuous Quench Flow System
The first 20 seconds of ozone decomposition in natural and
wastewater may be time-resolved with stopped-flow systems. However,
ozone and DOM's aromatic moieties have similar absorption peaks,
extinction coefficients and concentrations and can therefore not be
differentiated based on direct UV measurement. Commercial quench
Continuous quench flow system 27
flow systems are available but operate in a discontinuous mode (i.e.
one-push syringes), which does not easily allow sampling of larger
volumes for post-experimental analysis. Cho et al. (22) used a "flow
injection analytical" system allowing better resolution than batch
experiments, unfortunately it is not clear what is the dead time of the
apparatus, and if mixing is complete prior to the first data point.
The logic of the Continuous Quench Flow System (CQFS) developed
for this study is shown in Figure 2.2. The solution to be ozonated is
delivered with the first pump, Pl5 to the first mixer, M1+2, and rapidly
mixed with ozone delivered by the second pump, P2, (dead time in
M1+2: 20 milliseconds).
Quenching Agent
indigosulfite
2-3
1
Water characterization Post-experimental analysis Photometer
[O3]o / DOC / Br / pH / UV2M/2B5 pCBA / phenol / Br03 600 nm / spectra
Figure 2.2. Logic of the Continuous Quench Flow System. Two step-
motor controlled double-syringe pumps, one delivering the oxidant (P2)and the other the solution to be oxidized (Pi) are set at constant flow
rates. The solutions are mixed in M1+2, flow through one of 8 loopswith various volumes (Li %) and the oxidant residual is quenched with a
reagent delivered by a third syringe pump (P3). The resulting solution
flows into a flow-through photometric cell and/or is collected for
subsequent analysis.
Test solution
natural water
DOM
phenolBr
Oxidant solution/
ozone /chlorine /bromine L
A: p,
M,.2
Hill
28 Chapter 2
The reacting mixture then passes through one of 8 loops, Li %, with
volumes of 0.016, 0.101, 0.203, 0.245, 0.490, 0.980, 1.936, 3.872 mL.
The residual oxidant in the mixture is quenched with a reagent
(e.g. mdigo) delivered by a third pump, P3, in a second mixer,
Mi2+3- Given a total flow of 660 mL/h and a total dead volume of
0.048 mL, the mixture exits the second mixer, M12+3, at times 0.35,
0.81, 1.37, 1.60, 2.94, 5.61, 10.82 and 21.38 seconds. By successively
selecting different loop sizes, instead of modifying flow rates,
reaction times can be varied without affecting the flow regime (i.e.
the Reynold's number). The mixture, then, flows into the flow-
through cell and absorbance can be measured. The mixture
effluent can also be collected for post-processmg analysis
(e.g. SEC, IC, GC, HPLC).
Figure 2.3 shows an ideal signal from an experiment with CQFS. Each
absorbance step represents one time step (one loop). The difference
with the blank (indigo and water), AA, gives the ozone concentration.
Absorbance is measured continuously during each time step. This
allows the calculation of a standard deviation that is representative of
the compounded mixing efficacies of M1+2 and M12+3 (Figure 2.4b).
A
blank
11s1 loop12"" loop f
AA=C03 x 20000
8 "loop>
Figure 2.3. Ideal signal from a CQFS experiment Each absorbance
step represents one time step (one loop) The difference in absorbance
with the blank, AA, divided by the extinction coefficient of indigo
(20'000 M 1cm 1) gives the residual ozone concentration
Continuous quench flow system 29
As mentioned above, CQFS consists of four key subsystems: pumps,
mixers, flow loops, and flow-through cells with online photometer.
Step-motor controlled double-syringe pumps were used to prevent
back-pressure vanabikty from affecting the flow rate accuracy.
Two Kronlab LDP-5 (Pi, P2) and a Kronlab LDP-23 (P3) were used
with precisions of 0.1% (of flow rate) and able to handle back-pressure
up to 5 x 106 N/m2. The efficacy of the mixers is crucial because given
the flow rates and diameter of the tubmg the best achievable Reynold's
number is 460, which is significantly smaller than what is required to
obtain fully turbulent flows (Re ~ lO'OOO). Mixers must also be as small
as possible to minimize the dead volume (limits the rapidity at which
the first data point can be acquired) and must be effective even with
asymmetrical flow conditions (i.e. flow rate from water mlet is 10 times
larger than from oxidant inlet). PEEK mixing tees (VICI Jour Research,
Inc.) of 4 uL were used and yielded very good results in this particular
setup. The 8 flow-loops were made of Teflon tubing and connected
with two Teflon coated 8-way Hamilton HVX plug valves. The flow-
through cells were 1 cm Hellma (750 uL), V2 cm Hellma (375 uL) or
1 cm micro volume Hellma (80 uL). In this configuration and with the
maximum flow rates, CQFS allows a first measurement after
115 milliseconds and the determination of first-order rate constants
k' < 5 s! (= k" < 105 M h 1 with [substrate] = 50 |iM).
2.4 Results and Discussion
2.4.1 Accuracy of CQFS
To ensure the system's measurement accuracy down to 350 milkseconds
(i.e. no appearance of measurement artefacts caused by incomplete
mixing), the kinetics of oxidation of phenol was measured and
compared to pubkshed values.
30 Chapter 2
The second-order rate constants for the reaction of ozone with phenol and
phenolate are 1.3 x 103 M is1 and 1.4 x 109 M is1, respectively (23).
The apparent rate constant at specific pH values can be calculated based on
the above values of k"03 and p-Kaphenoi = 9.9 (23). Experiments were
conducted at pH 2.25 with 500 uM phenol in excess of 50 uM ozone to
ensure pseudo first-order conditions. 200 mM 1-propanol was added to the
solution to scavenge any HO' generated. Figure 2.4a shows the decrease of
ozone as a function of time. The decolounsation of indigo was measured at
each time step, and the extracted decrease of ozone concentration was
perfectly exponential down to 350 miUiseconds, yielding k'03app= 1240 M h 1.
This represents a difference of - 6.8 % with the pubkshed rate constants
(k"o3app = 1331 Mis1 at pH 2.25), well within the cited error margin
of 15 % (23). Figure 2.4a also displays very good akgnment of the data
points down to 350 milkseconds and an intercept at 0, indicating that CQFS
is not mixing-limited (i.e. no shoulder effect). The standard deviation of
absorbance was calculated for each time step. Figure 2.4b shows the
coefficient of vanation of absorbance (stdev(A)/Aavemge, n > 50) as a
function of reaction time, dearly as reaction time decreases, deviation from
the average absorption value increases, indicating that mixing becomes less
complete, but even at 350 ms the coefficient of vanation does not exceed 1.4%.
The above experiment was reversed with 53 uM ozone in excess of
1 uM phenol and pH was ad|usted to 4.15. 10 mM t-butanol was added as
HO' scavenger. The reaction between ozone and phenol was stopped by
quenching ozone with thiosulfate, and phenol was measured using HPLC
with fluorescence detection. The decrease in phenol concentration was
perfectly exponential over the measured time range (350 milkseconds to
20 seconds, data not shown), yielding k'03app
= 2100 M h \ a factor of
0.55 off the 3790 M h1 that is expected if ozone concentration decrease
is measured (23). This ratio corresponds to the stoichiometric factor of
the reaction between ozone and phenol (in moles of phenol per mole of
ozone) and is close to the pubkshed value of 0.48 at pH 7 (24).
Continuous quench flow system 31
Time [s]
O
Ö
a)
Hoigneetal 1983
This study
1 5% -
b)
^
Xl
^1 0% -
g
0 5% -
X
x^~—^_
0 0% -
Time [s]
Figure 2.4. Reaction of ozone with phenol (a) Open squares show
the measured decrease of 50 uM ozone induced by reaction with
500 uM phenol at pH 2 25, in presence of 200 mM 1-propanol as
HO" scavenger The solid line shows the decrease predicted with
published rate constants (23) (b) Coefficient of variation of
absorbance as a function of reaction time
32 Chapter 2
2.4.2 Reproducibility of measurements with CQFS
Figure 2.5 shows results of triplicate experiments, where ozone and
pCBA concentrations were measured during the initial 20 seconds
of ozone reaction with Opfikon wastewater. In Figure 2.5a,
90% confidence intervals (dashed lines) are on average 11% off the
average concentrations for the 55 uM ozone dose data series
(crosses) and 20% (not shown) for the 31 uM data series (open
symbols). The increase m the confidence interval at smaller ozone
doses is due to uncertainties linked with the measurement of small
ozone concentrations.
Similarly to ozone concentration measurements, measurement of
HO'-probe pCBA shows a good reproducibility: 90% confidence
intervals (dashed lines) are on average 5% off the average pCBA
values for the four data series shown here. Figures 2.5 includes
results from experiments performed with the same wastewater at
various times of the investigation (over ~50 days). Effects of agmg
are therefore compounded m the above confidence intervals.
Continuous quench flow system 33
Ozone Exposure [Ms]
0 E+00 2 E-05 4 E-05 6 E-05
20 25
3 E-05 1 E-04
I
\\
1
\
\
\
\
b)
*x~~~-__---
+
-_..£
""""---..__X
+
Figure 2.5. Reproducibility of ozonation experiments in Opfikonwastewater at pH 8 and ozone doses of 55 uM (2 6 mg/L) and
31 uM (1 5 mg/L) (a) Ozone decomposition as measured with CQFSfor triplicate experiments Dashed lines represent the 90% confidence
interval for the 55 uM series (b) HO"-induced oxidation of pCBA as a
function of ozone exposure Dashed lines represent the 90%
confidence interval
34 Chapter 2
2.4.3 Agreement between CQFS and batch experiments
To verify the complementanties of the methods, some expenments
were performed with both, a batch and a continuous quenched flow
system. Figure 2.6a shows the decrease of ozone concentration in Lake
Zurich water and display good agreement between CQFS (circles) and
batch (squares). A perfect alignment is difficult due to the way time is
measured in batch systems. The time needed to m|ect aqueous ozone
and obtain a homogeneous solution in a 500 ml bottle adds uncertainty
to the timing of data points below 60 seconds.
As demonstrated by the mset in Figure 2.6a, CQFS can easily time-
resolve the "instantaneous" ozone demand (IOD) in Lake Zurich
water. This shows that there is no "disappearance" of ozone during the
IOD but merely a decomposition that is too rapid to be measured with
a batch system. When displayed on a semi-log plot (i.e. ln([03]/[03]o vs.
time), the data points in Figure 2.6a do not line up in a straight line
(data not shown). This indicates that ozone decomposition in the first
20 seconds is mechanistically different than in the minute range where
it follows apparent first-order kinetics (25).
Figure 2.6b shows the change of Ret over time (Rct = J[HO']dt/J[03]dt:ratio of HO' to O3 exposure). Dunng the first 200 seconds,
R« decreases by two orders of magnitude from 2 x 10 6 —which is high
even for O3/H2O2 advanced oxidation processes (26)— to 3 x 10 8—
which is typical for ozonation of Lake Zurich water in the minute range
(27). During the first 100 seconds, Ret can be well fitted with a power
function; this was observed throughout our study in all waters and under
all conditions investigated (25). When it reaches the minute range,
however, R«becomes constant. This property of R« (i.e. dRct/dt = 0) is
well documented and is the rationale for its use as a key parameter to
model ozonation processes in drinking water treatment (15,27).
Continuous quench flow system 35
~v._
ci
1200300 600 900
Time [s]
b)
oE+oo -Uxrocc^r
0 1 1 10 100 1000
Time [s]
Figure 2.6. Comparison between CQFS (circles) and batch
experiments (squares) in Lake Zurich water at pH 8 and an ozone dose
of 50 uM (2 4 mg/L) (a) Ozone concentration as a function of time
(b) Rct (J |HO"]dt / J [03]dt) as a function of time on a log-log plot Topinset HO" exposure as a function of time Bottom inset O3 exposure
as a function of time
36 Chapter 2
In Figure 2.6b, however, it is interesting to note that R« is only constant
over one order of magnitude of time. In Lake Zurich water, this represents
a singularity in the entire kinetics history of ozone decomposition, which
covers four orders of magnitude of time (from ~ 0.2 to ~ 2000 seconds).
Insets in Figure 2.6b also show that while only roughly 8% of the total
ozone exposure occurs during the first 20 seconds, 25% of the total HO'
exposure occurs in the same period of time.
Figure 2.7a shows a comparison between CQFS and batch measurements
in wastewater effluent (Opfikon at pH 8). Given the high DOC
concentration, ozone reacted rapidly and could no longer be detected after
2 minutes. In contrast to experiments with natural water (Figure 2.6),
CQFS can only resolve part of the ozone decomposition in wastewater:
50% reacts prior to 350 milkseconds. Similarly to Lake Zurich water
experiments, the first values of R« obtained in Opfikon wastewater are of
the order of 10 6 (inset). However, in Lake Zurich water the first Ret value
is obtained when only 4% of the added ozone has reacted, in wastewater
the first R« value is obtained when 50% has already reacted. R« is
therefore likely to be significantly larger during the first milkseconds
following ozone addition to wastewater. Also, while R« = ~108,
100 seconds after ozone addition to lake Zurich water, it is one order of
magnitude larger (~10 ^ 100 seconds after addition to wastewater.
Figure 2.7b shows the increase of ozone exposure over time in
Opfikon wastewater. The sokd black squares show the curve obtained
when ozone exposure obtained from a CQFS experiment is added to
the first data point of the batch experiment. The open squares
exemplify the error that is made if the decrease in ozone concentration
is assumed to be linear (grey kne in Figure 2.7a, approximation in the
absence of CQFS data). This latter approximation overestimates initial
ozone exposure by roughly a factor of 2.
Continuous quench flow system
Time [s]
Figure 2.7. Comparison between CQFS (circles) and batch experiments
(squares) in Opfikon wastewater effluent at pH 8 and ozone dose of
4 mg/L (83 uM) (a) Ozone concentration decrease. Inset:
Corresponding R^ as a function of time (b) Ozone exposure as a
function of time. Assuming a linear decrease in ozone concentration
prior to the first batch measurement leads to overestimated exposures
(open squares).
38 Chapter 2
2.4.4 Oxidation and disinfection during wastewater ozonation
Based on eq 2 and the measured values of J03dt and jHO'dt, it should be
possible to model the degree of oxidation of any compound in wastewater
(assuming k"o3 and k"Ho- are known). To test this hypothesis, Opfikon and
Berlin wastewater effluent, Lake Zunch water and Berlin drinking water
were spiked with carbamazepine pnor to ozone addition. Figure 2.8 shows
a companson of measured (y-axis) and modeled (x-axis) carbamazepine
degradation. Symbols located on the dashed line show perfect agreement
between measurements and predictions. Sokd or black symbols represent
predictions that take both JHO'dt and JOsdt into account, while open or
grey symbols represent predictions where JHO'dt are neglected.
5% 10% 15%
Modeled C/C0 [-]
Figure 2.8. Measured versus modeled carbamazepine (Ca) oxidation
during the first 20 seconds of ozonation in various waters at pH 8
Black symbols prediction based on J03dt and JHO'dt, open symbols
prediction based only on J03dt Circles Berlin WW (2 3 mg03/L,2uMCa), star Opfikon WW (2 4 mg03/L, 1 uM Ca), squares and
triangles Berlin WW (12 mg03/L, 1 uM Ca), cross Berlin DW
(2 4 mg03/L, 1 uM Ca), diamond Zurich LW (2 4 mg03/L, 0 5 uM Ca)
Continuous quench flow system 39
Figure 2.8 clearly demonstrates that (1) eq 2 and second-order rate
constants determined in pure solutions are perfectly adequate to predict
the degree of oxidation during wastewater and natural water ozonation
and (il) HO' induced oxidations play a crucial role in the process and
shall not be omitted if accurate predictions are sought. The latter
comment is consistent with earlier findings in natural waters (28,29).
Given the demonstrated accuracy of the model (Figure 2.8), it is
interesting to investigate the degree of oxidation that can be predicted
for pharmaceutical compounds with various reactivities towards ozone.
Table 2.2 ksts the chosen compounds and their respective second-order
rate constants.
Table 2.2. Second-order rate constants of pharmaceuticals oxidation and
pathogens inactivation at pH 8 and T = 20 °C (9,30)
kraal's !] k ho- [M % i] k 03 [M % i]
17a ethmylestradiol 3 16 x 107 9 8 x 10' E coli 1 04 x 105
Sulfamethoxazol 2 5 x 106 5 5 x 10' Rotavirus 6 x 104
Diclofenac 106 7 5 x 10' Giardia lamblia cystsa 2 3 x 104
Carbamazepine 3 x 10= 8 8 x 10' Giardia murn cystsa 1 2 x 104
Bezafibrate 5 9 x 102 74 x 10' C parvum oocysts 6 7 x 102
Ibuprofen 96 74 x 109
Iopromid 8 x 10 ! 3 3 x 109
Diazepam 75 x 101 72 x 109 » at 25 °C
In Figure 2.9a the prediction is based on JHO'dt and JOsdt measured
for an ozone dose of 1.5 mg/L in Opfikon wastewater
(= 0.3 mgOs/mgDOC) at pH 8 and 20 seconds (at t = 20s: [03] = 0).
Compounds with k"o3 > 104 M H 1are completely (> 99%) oxidized by
ozone. Conversely, compounds with k'03 < 104 M h lare only partially
oxidized and almost exclusively by HO'. For example, although
diazepam (tranquikzer) and iopromid (contrast media) have roughly the
same k'03, diazepam's kMHo is twice as large as îopromid's inducing
almost twice as much degradation of diazepam.
40 Chapter 2
k"m < 104 M 's 1k"m > 104 M 's 1
3) Ozone Dose 1.5 mg/L
H HO'
k"03 > 104 M 's '
Figure 2.9. Pharmaceuticals oxidation in Opfikon wastewater at pH 8,
modeled with JHO'dt and JOsdt measured over 20 seconds. Light grey
bars: compounds fractions oxidized by HO", dark grey bars: compoundsfractions oxidized by O3. (a) O3 dose = 1.5 mg/L. (b) O3 dose = 4 mg/L.
Continuous quench flow system 41
In Figure 2.9b, O3 dose is increased to 4 mg/L, which results in a
similar trend as in Figure 2.9a. However, while the degree of
HO'-mduced oxidation is increased significantly at higher ozone doses,
that of 03-mduced oxidation is not. For both doses, it must be noted
that for compounds with k'03 < 104 M hl, the process is
undistinguishable from an advanced oxidation process (AOP).
Figure 2.9 is also important from the standpoint of product formation.
Compounds with k"o3 > 104 M h 1 will generate mostly 03-mduced
metabolites. These have been shown to often be bio-chemically
inactive (11). In contrast, compounds with k'03 < 104 M h l will generate
mostly HO'-induced metaboktes about which kttle is known (10).
In Figure 2.10, the degrees of inactivation of Cryptosporidiumparvum
oocysts, Giardia lamblia cysts, Giardia muns cysts, Rotavirus and E. coll
were modeled using eq 3, previously measured JOsdt and the rate
constants shown in Table 2.2. Clearly, ozone exposure in wastewater
is small and 03-resistant microorganisms such as C. parvum are not
inactivated. However, at a moderate ozone dose of 2.5 mg/L
(= 0.55 mgOs/mgDOC), 90% inactivation of Giardia Muns, 5 logs
inactivation of Rotavirus and more than 6 logs inactivation of E. coll
are predicted. This calculation demonstrates that a significant
disinfection potential is available even though ozone has entirely
reacted prior to 20 seconds. It must be noted, however, that the
above model does not take particle shielding or reactor dynamics into
consideration which would impact the process efficiency (31).
42 Chapter 2
Figure 2.10. Modeled inactivations of microorganisms based on JC^dtmeasured in Opfikon wastewater at pH 8 and 20 seconds for ozone
doses of 1 5, 2 5 and 4 mg/L
The importance and challenges in quantifying JOsdt and JHO'dt during
wastewater ozonation have been demonstrated. Although CQFS is a
promising research tool, in its present form it is kmited to bench scale
investigations. As an alternative, the development of oxidant probe
compounds to back-calculate JOsdt and JHO'dt might be a promising
idea —akm to the biodosimetry concept in UV disinfection. However,
as for UV systems, imperfect reactor dynamics could significantly
impair the concept (32,33) and may explain discrepancies observed
earlier. During pilot experiments, Huber et al. (2005) used the partial
oxidation of some compounds to back-calculate JHO'dt and JOsdt and
predict the degradation of other compounds. Predictions did not
always agree with measurements. Three hypotheses were formulated to
explain the discrepancies: (i) gaz-kquid mass transfer kmitation,
Continuous quench flow system 43
(11) adsorption of compounds onto wastewater particles and
(in) mapphcabikty to wastewater systems of second-order rate
constants obtained in pure water expenments in the laboratory.
The first two hypotheses were previously reacted on the base of
calculations (12) and the third one can be reacted on the base of the
present study. As mentioned above, the origin of the discrepancies may
ke in the assumptions that must be made when back-calculating oxidant
exposures (reactor dynamics). Further research with regard to probe
compounds is therefore warranted prior to encouraging their broader
use in the industry.
2.5 Conclusions
• Experiments in Lake Zunch water demonstrated that what appears as
"instantaneous" ozone demand (IOD) with batch systems can be
entirely time-resolved with CQFS. Hence, IOD is an operational
parameter that is not related to the chemical mechanisms taking place.
In contrast, the terms initialphase and secondphase are encouraged as
they have mechanistic definitions. In the minutes range, the secondphase
of ozone decomposition follows an empirical first-order rate law.
During the initialphase, however, ozone decomposition follows higher-
order kinetics indicating that other reaction mechanisms might be
important (for further discussion see (25,34)).
• During the initialphase of Lake Zurich water ozonation and during
Opfikon wastewater ozonation, R^ is 2 to 3 orders of magnitude
larger than dunng the second phase of natural water ozonation.
Most H2O2/O3 AOPs do not reach such large R« values (26).
This, de facto, places wastewater ozonation in the same category as
ozone based advanced oxidation processes (AOP).
44 Chapter 2
• In natural water, although R« is not constant for most of the
kinetics history of ozone decomposition, it does become constant
in the minute time-range. Hence, the R« concept has been
successfully applied to natural water ozonation and offers great
simplification for modeling purposes. In wastewater ozonation,
however, it is not recommended to use the R« concept to model
compounds degradation as Rct is never constant during the
process. Notwithstanding this limitation, Rct remains an essential
parameter for all ozonation-based treatments because it situates
the processes on a scale from mostly 03-driven to mostly HO'-
dnven oxidation mechanisms.
• There is significant ozone exposure "hidden" in the first 20 seconds
of ozonation. Not accounting for it leads to overly-conservative
assumptions. Although no ozone residual is left 20 seconds after
2.5 mg03/L addition to Opfikon wastewater, ozone exposure is
large enough to inactivate more than 6 logs E. coli.
• Ozonation of wastewaters degrade pharmaceuticals with a high
efficacy. Earker results can be well confirmed and explained by the
present study: compounds with high ozone reactivity (> 104 M H *)
are readily oxidized by ozone, while those with lower ozone
reactivities are mostly oxidized by HO'. This might have an
important impact on product formation and bio-chemical
inactivation of the pharmaceutical compounds.
2.6 Acknowledgments
We thank CIRSEE - Suez Environnement for financial support;
Isabelle Baudrn, Auguste Bruchet, Zdravka Do-Quang, Mane-Laure
Janex, Jean-Michel Lamé and Philippe Savoye (CIRSEE) for fruitful
Continuous quench flow system 45
discussions; Michael Dodd, Marc Huber and Gretchen Onstad for
insightful comments. Special thanks to Adnano Joss for his insight on
the treatment processes involved at the Opfikon WWTP and to Patnce
Goosse and Sebastian Zabczynski for their help in obtaining the water
samples. Moreover, we thank the German Ministry of Education and
Research (BMBF) for supporting the research stay of Jochen
Schumacher at eawag.
2.7 References
1 Ternes, T A,Occurence of drugs in german sewage treatment plants and
rivers Wat. Res. 1998, 32, 3245-3260
2 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 Environ. Sa. Technol. 2002, 36, 1202-1211
3 Thorpe, K L, Cummings, R I, Hutchinson, T H , Scholze, M , Bnghty, G,
Sumpter, J P, Tyler, C R, Relative Potencies and Combination Effects of
Steroidal Estrogens in Fish Environ. Sa. Technol. 2003, 37,1142-1149
4 Purdom, C E, Hardiman, P A, Bye, V J, Eno, N C, Tyler, C R,
Sumpter, J P, Estrogenic effects of effluents from sewage treatment
works Chem. Ecol. 1994, 8, 275-285
5 Schwartz, T, Kohnen, W, Jansen, B, Obst, U, Detection of antibiotic-
resistant bacteria and their resistance genes in wastewater, surface water,
and drinking water biofilms FEMS Microbiol. Ecol. 2003, 43, 325-335
6 Goni-Urnza, M, Capdepuy, M, Arpin, C, Raymond, N, Caumette, P,
Quentin, C, Impact of an urban effluent on antibiotic resistance of nvenne
Enterobactenaceae and Aeromonas spp Appl. Env. Microb. 2000, 66, 125-132
7 Ferrari, B, Paxeus, N, Lo Giudice, R, Pollio, A, Garnc, J,
Ecotoxicological impact of pharmaceuticals found in treated wastewaters
study of carbamazepine, clofibric acid, and diclofenac Ecotox. Environ.
Safe. 2003, 55, 359-370
8 Laville, N, Ait-Aissa, S, Gomez, E, Casellas, C, Porcher, J M ,
Effects
of human pharmaceuticals on cytotoxicity, EROD activity and ROS
production in fish hepatocytes Toxicology 2004, 196, 41-55
9 Huber, M M, Canonica, S, Park, G-Y, von Gunten, U, Oxidation of
pharmaceuticals during ozonation and advanced oxidation processes
Environ. Sa. Technol. 2003, 37, 1016-1024
46 Chapter 2
10 Dodd, M C, Buffle, M-O, von Gunten, U, Moiety-specific oxidation of
antibacterial molecules by aqueous ozone Reaction kinetics and relevance
to ozone-based wastewater treatment Environ. Sei. Technol., in press, 2006
11 Huber, M M, Ternes, T A, von Gunten, U, Removal of Estrogenic
Activity and Formation of Oxidation Products during Ozonation of 17a-
Ethinylestradiol Environ. Sei. Technol. 2004, 38, 5177-5186
12 Huber, M M, Gobel, A , Joss, A , Hermann, N , Loeffler, D, McArdell,
C S, Ried, A, Siegrist, H, Ternes, T, von Gunten, U, Oxidation of
pharmaceuticals during ozonation of municipal wastewater effluents a
pilot study Environ. Sei. Technol. 2005, in Press
13 Hoigné, J, Characterization of water quality critena for ozonation processes
Parti Minimal set of analytical data O^one Sa. Eng 1994, 16, 113-120
14 Hoigné, J, Bader, H, Characterization of water quality criteria for
ozonation processes Part2 Lifetime of added ozone O^one Sei. Eng.1994, 16, 121-134
15 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios duringozonation processes I The Rct concept O^one Sei. Eng. 1999, 21, 239-260
16 Rakness, K L, Corsaro, K M, Hale, G, Blank, B D, Watewater
disinfection with ozone - Process control and operating results O^one Sei.
Eng. 1993, 15, 497-514
17 Paraskeva, P, Graham, N J D, Ozonation of municipal wastewater
effluents Wat. Envir. Res. 2002, 74, 569-581
18 Buffle, M-O, Galli, S, von Gunten, U, Enhanced Bromate Control
during Ozonation The Chlorine-Ammonia Process Environ. Sei. Technol.
2004, i£, 5187-5195
19 Joss, A , Andersen, H , Ternes, T, Richie, P R, Siegrist, H ,Removal of
Estrogens in Municipal Wastewater Treatment under Aerobic and
Anaerobic Conditions Consequences for Plant Optimization Environ. Sei.
Technol. 2004, 38, 3047-3055
20 Schumacher, J, Pi, Y Z, Jekel, M ,
Ozonation of persistent DOC in
municipal WWTP effluent for groundwater recharge Water Sei. Technol.
2004,42,305-310
21 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Wat. Res. 1981, 15, 449-456
22 Cho, M, Kim, H , Cho, S H, Yoon, J, Investigation of ozone reaction in nver
waters causing instantaneous ozone demand O^tne Sei. Eng 2003, 25, 251-259
23 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water - II Dissociating organic compoundsWat. Res. 1983, 17, 185-194
24 Mvula, E, von Sonntag, C, Ozonolysis of phenols in aqueous solution
Org. Biomo/. Chem. 2003, /, 1749-1756
Continuous quench flow system 47
25 Buffle, M-O, Schumacher, J, Meylan, S, Jekel, M, von Gunten, U,
Ozonation and Advanced Oxidation of Wastewater Effect of O3 Dose,
pH, DOM and HO'-scavengers on Ozone Decomposition and HO"
Generation O^one Sei. Eng. accepted, 2006
26 Acero, J L, von Gunten, U, Characterization of oxidation processes
ozonation and the AOP 03/H202 Jour. AWWA 2001, 93, 99-100
27 Elovitz, M S, von Gunten, U, Kaiser, H-P, Hydroxyl radical/ozone
ratios during ozonation processes II The effect of temperature, pH,
alkalinity and DOM properties O^one Sei. Eng. 2000, 22, 123-150
28 Acero, J L, Haderlein, S B
, Schmidt, T C, Suter, M J -F, von Gunten,
U, MTBE oxidation by conventional ozonation and the combination
ozone/hydrogen peroxide Efficiency of the process and bromate
formation Environ. Sei. Technol. 2001, 35, 4252-4259
29 Acero, J L, Stemmler, K, von Gunten, U, Degradation kinetics of atrazine
and its degradation products with ozone and OH radicals A predictive tool
for drinking water treatment Environ. Sei. Technol. 2000, 34, 591-597
30 von Gunten, U, Ozonation of drinking water Part II Disinfection and
by-product formation in presence of bromide, iodide and chlorine Wat.
Res. 2003, 37, 1469-1487
31 Xu, P, Janex, M-L, Savoye, P, Cockx, A, Lazarova, V, Wastewater
disinfection by ozone main parameters for process design Wat. Res.
2002, 36, 1043-1055
32 Mackey, E D, Hargy, T M , Wright, H B, Malley, J P, Jr , Cushing, R S
,
Treatment technologies Comparing Cryptosporidium and MS2 bioassays-
lmphcations for UV reactor validation Jour. AWWA 2002, 94, 62-69
33 Buffle, M -O, Chiu, K -p, Taghipour, F UV Reactor Conceptualization and
Performance Optimization with Computer Modeling New Orleans, LA, USA
34 Buffle, M -O, von Gunten, U, Phenol and Amine-Induced HO"
Generation During the Initial Phase of Natural Water Ozonation
Environ. Sei. Technol., accepted, 2006
3 Ozonation and Advanced Oxidation of
Wastewater: Effect of O3 Dose, pH, DOM
and HO'-scavengers on Ozone
Decomposition and HO' Generation
Marc-Olivier Buffle, Jochen Schumacher,
Sébastien Meylan, Martin Jekel, Urs von Gunten,
O^one Science andEngineering 2006
3.1 Abstract
The decomposition of ozpne in wastewater is observed starting 350 milliseconds
after ozpne addition. It seems not to be controlled by the autocatalytic chain reaction,
but rather by direct reactions with reactive moieties of the dissolved organic matter
(DOM). A larger ozpne dose increases ozpne consumptionprior to 350 milliseconds
but decreases the rate of ozpne decomposition later on; this effect is preàcted by a
second-order kinetic model. Transferred Ozpne Dose (TOD) is poorly correlated
with ozpne exposure (— jfOJdt) indicating that TOD is not a suitable parameter
for the prediction of disinfection or oxidation in wastewater. HO' concentrations
(> ia'° M) and Ra (= l[HO']dt/l[OJdt > 106) are larger than in most
advanced oxidation processes (AOP), but rapidly decrease over time. Ra also
decreases with increasing pre-ozpnation doses. An increase in pH accelerates ozpne
decomposition and HO' generation; this effect ispredicted by a kinetic model taking
into account deprotonation of reactive moieties of the DOM. DOC emerges as a
cruaal water qualityparameter that might be of use to normalize ozpne doses when
comparing ozonation in different wastewaters. A rapid drop of absorbance in the
water matrix —with a maximum between 255-285 nm— is noticeable in thefirst
350 milliseconds and is directly proportional to ozpne consumption. The rate of
absorbance decrease at 285 nm is first order with respect to ozpne concentration.
A kinetic model is introduced to explore ozpne decomposition induced by
distributions of reactive moieties at sub-stoichiometnc ozpne concentrations. The
model helps visualize and comprehend the operationally-defined "instantaneous ozpne
demand" observed during ozpne batch experiments with DOM-containing waters.
50 Chapter 3
3.2 Introduction
While the use of ozone for wastewater disinfection goes back to the
1970's (1) wastewater ozonation has received renewed attention with
the discovery of ozone's ability to efficiently degrade certain classes of
pharmaceutical compounds (2-6).
Although much effort has been invested in trying to characterize
wastewater ozonation, the process has long remained a black box.
Studies have mostly remained empirical with the optimization of
operating process parameters being based on a single end-point
—e.g. bacterial plate counts (7,8). Lately, researchers have tried to
describe the process with more meaningful empirical parameters such
as the transferred ozone dose (TOD: ozone consumed by the water
matrix). However, results show that oxidation and disinfection
predictions based on TOD are difficult (9-12).
Ideally, the extent of oxidation of a compound, P, during wastewater
ozonation in a well mixed reactor can be accurately predicted with eq 1 (6).
-k'o, l[03] dt-k"HO \[HO-]dt[P] = [P]0-e
' ° with [Oj= fit) [HO] =g(t) (1)
To solve eq 1, however, second order rate constants (k"0i, k"H0.) and
oxidants exposures (J[OJ<Ä , j[HO"\dt) must be known (for
disinfection, HO' terms are usually neglected). While k" are physical-
chemical constants and need be determined only once, oxidant
exposures are functions of a number of operating and environmental
conditions such as the ozone dose, scavenging capacity of the
wastewater matrix and pH. Oxidant exposures are therefore different
for each wastewater and cannot be predicted or easily measured,
Characterization of ozone decomposition in wastewater 51
hence, the ongoing effort (e.g. TOD) to find alternative parameters
allowing for the prediction of oxidation in wastewater.
In a recent study, the degrees of oxidation of probe compounds during
pilot scale ozonation were used to back-calculate O3 and HO' exposures
(using eq 1). However, potential shortcomings warrant further
investigation prior to broader appkcation of the concept (6,13).
Moreover, while the information gained might be useful in predicting the
degree of oxidation of specific compounds, it is cumulative and the
dynamics of the system cannot be extracted. It is therefore necessary to
introduce a new method to gain insight into the dynamics of the system.
We showed earker that the use of a continuous quench flow
system (CQFS) allowed the measurement of ozone decomposition
over sub-second time-scales (6). With CQFS, ozone decomposition in
wastewater can be time-resolved and impacts of operational and water
quakty parameters can now be investigated.
3.2.1 Theoretical background
In natural waters, the second phase of ozone decomposition (t > ~20 s)
can be empirically modelled with an apparent first-order rate law.
This second phase has been extensively studied for natural waters and is
attributed to radical-type chain reactions during which HO' radicals are
formed as secondary oxidants (14,15). Descnption of oxidation
mechanisms during an ozonation process must therefore take both, HO'
and O3 into account (see eq 1). The relative importance of HO'- versus
03-based oxidations can be quantified with R«, the ratio of HO' exposure
to O3 exposure (J[HO']dt / J[Os]dt). The R« concept was developed as a
tool to model oxidation of compounds during natural water ozonation
following the observation that R« remains constant during the second
phase —i.e. during most of the ozonation process in natural waters (16).
52 Chapter 3
During the initial phase (t < ~20s) of natural water ozonation
(operationally defined as "instantaneous ozone demand"), ozone
decomposition is more rapid and does not follow a first order rate law
as in the second phase (6). Using CQFS, it was also recently shown that
R« values are very high and decrease exponentially before reaching a
constant value in the second phase (6). The initiation step of the radical
chain reaction (HO + O3), however, has a small rate constant
(k" = 70 M H h ti/2 = 9860 s at pH 8) and cannot be expected to play an
important role during the initial phase. Hence, other mechanisms, such
as bimolecular reactions between ozone and specific moieties of the
dissolved organic matter (DOM), might be responsible for initiating the
decomposition of ozone and generating HO' during the initial phase
of natural water ozonation. For example, phenokc and ammo groups,
ubiquitous in DOM, react readily with ozone when deprotonated, and
have been shown to partially generate O3', which readily decomposes
to HO' (for an extensive discussion see (17)).
In wastewater ozonation, it was recently shown that Rct values and HO'
exposures are very high and decrease exponentially during the entire
duration of ozone decomposition (6). Hence, it seems that ozone
decomposition kinetics m wastewater is rather analogous to the initial
phase of ozone decomposition m natural water. One might therefore
hypothesise that the main mechanisms of ozone decomposition in
wastewater are its direct reactions with specific reactive moieties of the
DOM (phenokc, ammo and olefinic groups).
In this article we investigate the effect of ozone dose, pre-ozonation, pH
and DOM on the kinetics of ozone decomposition and HO' formation
in wastewater, starting 350 milkseconds after ozone addition. The effect
of ozonation on the absorption spectrum of wastewater is quantified.
Finally, we introduce an exploratory model that helps grasp the type of
processes that might be taking place during wastewater ozonation.
Characterization of ozone decomposition in wastewater 53
3.3 Materials and Methods
3.3.1 Reagents.
All reagents were of analytical grade. All solutions were prepared with
MilkQ water with a resistivity> 18 MDtm. Ozone stock solutions
were prepared as described elsewhere (18), concentration was typically
1.6 mM (77 mgOs/L).
3.3.2 Water characteristics
Waters were buffered with 0.5 mM borate for all experiments at pH 8
and 0.5 mM phosphate for lower pH and ad|usted with NaOH or
H2SO4, respectively. pH was venfied at the beginning and end of each
experiment and stayed constant (+ 0.05 pH unit). All waters
(see Table 3.1) were filtered at 0.45 urn (cellulose nitrate) and stored at
5°C for the entire duration of the investigation. All experiments were
performed at room temperature (22 ± 1°Q. The Opfikon wastewater
treatment plant (Zurich, Switzerland) is described elsewhere (19);
the water was obtained post sand filtration. Berkn wastewater was
obtained fiom the effluent of a secondary treatment train at the
Ruhleben WWTP, Germany (20). To compare the effect of different
DOM origins on ozone decomposition, a batch of Berkn wastewater
was diluted 1:1 with nanopure water, thus approaching Opfikon
wastewater DOC concentrations. Lake Zurich water was collected fiom
the raw water intake of Zurich drinking water treatment plant,
30 meters below the surface of the lake. In experiments where HO'
had to be scavenged, either 12 mM 1-propanol
(k"Ho. = 2.8 x 109 M h !) or 12 mM tert-butanol (k"HO- = 6 x 108 M h !)
were used, inducing HO' scavenging rates of 3.36 x 107 s1 and
7.2 x 106 s1, respectively.
54 Chapter 3
Table 3.1. Water quakty parameters of the tested waters.
DOC ^* ~3 NH3/NH4+~
Alkalm
~
mgC/L a285nm mgN/L usN/L MgN/L .a'"' P.
WatertyPefrxMQ mi U fuM) T^M) <mM> »
Opfikon WW (34?55) ^ ^ 330(24) ^ 36 79
B—W ^^
^ 52(4) (fß) 34 BO
Zunch LW (1\47) 3 1 5(0 36) ^ 2 4 7 8
WW wastewater effluent, LW lake water
3.3.3 Methods
HO' exposure was back-calculated using the oxidation of
para-chlorobenzoic acid (pCBA) analyzed with HPLC (16). Ozone
was measured online with a Vanan Cary 100, either directly at 258 nm
(e = 3000 M icma) or with the mdigo reagent at 600 nm
(e = 20'000 M !cm *) (21). The water matrix absorption spectra were
measured online with a Vanan Cary 100 using 6 mM sulphite as
quenching solution. The change in absorption was measured at
285 nm because it was close to the spectrum difference maximum
measured and is used as a more selective wavelength to measure
aromatics in natural waters (22). Size Exclusion Chromatography with
online UV detector, organic carbon detector and organic nitrogen
detector —LC-OCD-OND— (DOC-Labor Dr.Huber, Germany)
was used to determine the effect of ozone on DOM fractions (23).
3.3.4 Continuous Quench Flow System (CQFS)
The system has been descnbed and charactenzed previously (6).
With CQFS, aqueous ozone is continuously mixed with wastewater,
the reaction takes place in one of 8 flow loops of vanous volumes
and stopped at the outlet of the loops where mixing with a quenching
agent occurs. CQFS allows a first ozone concentration measurement
Characterization of ozone decomposition in wastewater 55
115 milliseconds after initial contact with ozone and the
determination of second-order rate constants k'03 < 105 M h 1. 90%
confidence intervals of ozone concentration measurements in
wastewater (for O3 dose of 55 uM) were on average 11% off the
mean values (n = 3). The HO'-probe pCBA displayed 90%
confidence intervals on average 5% off the mean values (n = 4) (6).
Examples of tnphcate expenments are shown in Figure 3.1a.
3.3.5 Modeling
ACUCHEM is a program for solving systems of coupled
differential equations describing the temporal behaviour of spatially
homogeneous, isothermal, multi-component chemical reaction (24).
It can handle up to 40 species and up to 80 simultaneous reactions.
ACUCHEM was used in this research to conceptually explore the
effect of reactive moieties distnbutions on ozone decomposition.
Vanous distnbutions were investigated using concentration ranges
and rate constants of environmental relevance. The second order
rate constants of reactions with ozone (n = 455) and HO'
(n = 1254) in aqueous solutions were extracted from the National
Standard database (25).
56 Chapter 3
3.4 Results and Discussion
3.4.1 Effects of Operating Parameters
3.4.1.1 Ozone Dose
Dose is a key control parameter in full-scale ozonation plants.
Therefore, it is important to understand its effect on observed ozone
decomposition kinetics. Experiments were conducted with
1.5,2,2.5 mg03/L (31, 41, 52 uM) added to Opfikon wastewater.
Varying the ozone dose changed the observed kinetics of ozone
decomposition, as can be seen in Figure 3.1a. While ozone
consumption pnor to 350 milliseconds increases with increasing dose,
the rate of ozone decomposition decreases (mset of Figure 3.1a).
At all doses tested here, no residual ozone could be measured beyond
20 seconds. As shown by the mset of Figure 3.1a, ozone
decomposition does not follow apparent first-order kinetics as it would
during the second phase of natural water ozonation.
Figure 3.1b shows that all ozone doses generate the same HO' exposure
(l[H01dt) at equal ozone exposures (J[OJ<i). Inset of Figure 3.1b shows
transient HO' concentrations. HO' concentration can be calculated by
taking the time differential of the HO' exposure time function (eq 3),
which is back-calculated fiom the oxidation of pCBA (eq 2):
j[HO']dt
JlpCBAl{[pCBA]0~
^HO-pCBA
with \pCBA] =f (t) (2)
=
djjlHO-w)wlth[H01=g^ (3)
dt
Characterization of ozone decomposition in wastewater 57
Time [s]
Ozone Exposure [Ms]1 E-04
m
O-O 25 -
5m
O
J '
b)
1 E10
J.1E1,
O
D HE 12
^Cà. ,E14
o
jPa c
4 8 12 1
Time[s]
O
O
6
Figure 3.1. Effect of ozone dose on ozone decomposition kinetics
(a) Ozone concentration versus time in Opfikon wastewater at pH 8 and
various ozone doses in tnplicate experiments (open circles 52 uM
(2 5 mg/L), solid triangles 41 uM (2 mg/L), open squares 31uM
(1 5 mg/L)) Solid lines predicted ozone decrease (see exploratory model)Inset same information on a log plot (b) Decrease of HO"-probe, pCBA,versus ozone exposure corresponding to the experiments in (a)Inset transient HO" concentrations versus time for ozone dose 1 5 mg/L
58 Chapter 3
HO' concentrations are very high —0.1400 nM (2.4 ngHO'/L) after
350 milliseconds— and are likely to be significantly higher earker in
the process. As a comparison, HO' concentrations in lab-scale
UV/H2O2 experiments with natural waters reach up to
0.0010 nM (26) and up to 0.0012 nM in an O3/H2O2 process at
pH 7 (27). The transient HO' concentrations increased with
increasing ozone dose at equal time and exposures (data not shown).
An important feature of the HO' concentration profiles is a decrease
by more than two orders of magnitude dunng the first seconds
following ozone addition.
As mentioned earker, the second phase of ozone decomposition in
natural water (t > 20 s) can be empirically modelled with an apparent
first-order rate constant (e.g. k' = 0.002 s1
in Lake Zurich water
at pH 8, (28)). As shown in Figure 3.2a, however, ozone
decomposition in wastewater cannot be modelled with a constant k'.
In Opfikon and Berlin wastewater effluents, the decreases of k' over
time could be fitted with power functions (k = oct ). The exponents
(ß) were similar for the various ozone doses and waters investigated,
but the curves shifted to higher values of k (a increased) as
normalized doses (mgOj/mgDOC) decreased (data not shown).
Characterization of ozone decomposition in wastewater 59
3.4.1.2 Transferred ozone dose
TOD has been proposed as a key parameter to charactenze wastewater
ozonation. It can be defined as the amount of aqueous ozone
consumed by the water matnx:
TOD = [03 ] 0- [03 ] with tR = residence time in contactor (4)
In full-scale wastewater contactor, aqueous ozone concentration at the
reactor outlet ([0J/=/R) is often nil so that TOD=[OJ/=0.
TOD can then be simply obtained from a mass balance calculation
based on the measurement of ozone in the gas phase.
Using CQFS to measure ozone exposures (J[Oj]dt in eq 1),
JOjdt was plotted against the corresponding calculated TOD for
36 experiments with Berlin (solid triangles) and Opfikon
(open circles) wastewater effluents (Figure 3.2b). There seems to be
little relationship between both parameters; ozone exposure can
vary more than a factor of two for the same transferred ozone
dose. TOD is therefore not appropnate to accurately predict the
resulting degree of oxidation or disinfection caused by ozone in
wastewater effluents. This likely explains difficulties linked with the
use of TOD in earlier studies. Moreover, TOD can clearly not be
used to extrapolate results from one type of water to another
(compare differences between Berlm-triangles and Opfikon-circles).
60 Chapter 3
a)
10 15
Time [s]
b) o
o
o
o
AA
A
o
o o°o
ooo
AA*A
AA
o
A
A
0A
A *
Transferred Ozone Dose [mg/L]
Figure 3.2. (a) Apparent first-order rate constant of ozone
decomposition, k1, in Opfikon wastewater at pH 8 and an ozone dose
of 52 uM (2.5 mg/L) (b) Total ozone exposure (= J[03]dt until ozone
has fully reacted) plotted against Transferred Ozone Dose for Berlin
(triangles, n = 17) and Opfikon (open circles, n = 19) wastewater
effluents at pH 8 and various ozone doses.
Characterization of ozone decomposition in wastewater 61
3.4.1.3 Pre-ozonation dose
In drinking water treatment a pre-ozonation step is often appked directly after
the raw water intake for colour removal, primary disinfection and/or to
improve the efficiency of subsequent flocculation processes. Although such
process trains are unlikely scenanos for wastewater treatment plants, it is
interesting to investigate ozone decomposition kinetics in wastewater that has
been pre-oxidized. Figure 3.3a shows ozone decomposition (2.5 mg/L O3
dose) in Opfikon wastewater following its pre-oxidation with vanous doses of
ozone (0,1,1.5,2.5 mg/L). After pre-oxidation with 2.5 mg/L (stars), ozone
decomposition can be almost entirely resolved with the continuous quench
flow system, suggesting that most fast reacting species have been oxidized.
In Figure 3.3b, R* values are plotted as a function of time on a double
loganthmic scale. The magnitude of Rct indicate that dunng the first
seconds of wastewater ozonation, the importance of HO'-based oxidation
is significant for all compounds with k'03 < 104 M h \
i.e. wastewater ozonation can be categonzed as an 03-based AOP.
For all Berlin and Opfikon wastewater experiments performed in this
investigation, R« decrease over time was well fitted by power functions with
very similar slopes —ß— (i.e. Rct= a• rt t = time [s]; a = f(pH, dose, DOQ;
and ßavg = 0.50 with a(ß) = 0.14, conf.int95%(ß) = [0.46; 0.54], n = 50).
The open symbols in the inset of Figure 3.3b show a cross section of the
main graph at 1 second (indicated by the vertical dashed line).
Clearly, increasing the pre-ozonation dose decreases Rct dunng subsequent
ozonation, however it levels off at higher doses. The top data set in the inset
of Figure 3.3b (sokd diamonds) is a cross section through the
corresponding experimental measurement obtained with Berlin wastewater.
Due to higher DOM concentration (DOCBedn = 1-9 x DOCopfikon) the
absolute value of Rct in Berlin wastewater is larger than in Opfikon
wastewater (RctBain ~ 2 x Rctopfton), but the relative change of Rct as a
function of pre-03 dose is very similar in both waters.
62 Chapter 3
1 »>- X -- 52 MM (2 5 mg/L)
1- B -- 31 MM (1 5 mg/L)
\ - Ä -
- G -
- 21 MM (1 mg/L)
-OmM
40 1 Q *H
* ta
i ^"
~Q-..
<3 a. "-~^
% ~
~A_~
B
Q
Q-.
"
~
-&_
-
_
^
~~
~
~o
n ,
Time [s]
b)
0 25 50 75 100
Pre oaanation DoseQjM]
Time [s]
Figure 3.3. Effect of various pre-ozonation doses (0, 1, 1 5,
2 5 mg03/L) on subsequent ozonation of Opfikon wastewater at pH 8
and 2 5 mgOs/L (a) Ozone concentration decrease as a function of
time (b) Rct plotted as a function of time on a double log-plot for the
same pre-ozonation doses as in (a) Inset Rct versus pre-ozonation
dose after 1 s (vertical line in main figure), solid diamonds show results
from experiments performed with Berlin wastewater
Characterization of ozone decomposition in wastewater 63
3.4.2 Effects of Water Quality Parameters
3.4.2.1 Role of DOC
Not merely the concentration of dissolved organic matter (quantified
as DOC) but also its composition (i.e. origin) might determine the
rate of decomposition of aqueous ozone. Hence, it is interesting to
compare waters of different ongms with the same DOC
concentrations or ozonated with similar O3 dose to DOC ratios.
Ozone decomposition was measured in Opfikon and Berlin
wastewaters as well as in 1:1 diluted Berkn wastewater. Figure 3.4a
shows ozone reacting much faster in Berlin (solid squares) than in
Opfikon wastewater (open circles), as is expected given the difference
in DOC concentrations (8.5 mg/L vs. 4.5 mg/L). Diluted Berlin
wastewater (solid triangles) however shows an ozone decomposition
profile that is strikingly similar to that of Opfikon wastewater.
Figure 3.4b shows HO'-mduced pCBA oxidation as a function of
ozone exposure for the three cases discussed above.
Berlin wastewater (sokd squares) shows a large HO' exposure at
350 milliseconds, but all ozone is decomposed rapidly. Diluted Berlin
wastewater (solid triangles) shows significantly larger HO' exposures
than Opfikon wastewater (circles). This is caused by a reduction of
the alkalinity following dilution. Carbonate is a key HO'-scavenger; its
decrease leads to a larger HO' exposure. Even though HO' exposure
(Figure 3.4b) is larger in diluted Berlin wastewater, its ozone
decomposition kinetics is similar to that of Opfikon wastewater
(Figure 3.4a). This is an indication that the autocatalytic chain reaction
may not be controlling ozone decomposition in these wastewaters.
64 Chapter 3
Time [s]3 4 8 1
-1 -
Aa)
__
Berlin 1 1
-r -2-
SOpfikon
Ô
£ -3-
-4-
Berlin
OE+00
11
Ozone Exposure [Ms]
1 E-04 2 E-04
b)
Opfikon
Figure 3.4. Experiments at pH 8 and an ozone dose of 2 5 mg/L
(52 uM) in Berlin wastewater (solid squares), Opfikon wastewater
(open circles) and 1 1 diluted Berlin wastewater (sohd triangles)
(a) Ozone decomposition kinetics (b) Oxidation of the HO"-probe
pCBA as a function of ozone exposure
Characterization of ozone decomposition in wastewater 65
It was shown earlier that ozone exposures do not correlate with
transferred ozone doses (Figure 3.2b). However, plotting ozone
exposures against DOC-normaksed O3 doses (O3/DOC in
mg03/mgC or molOs/molQ for experiments in Opfikon and Berkn
wastewaters seemed to akgn all results in a non-linear but unique
relationship (data not shown). Similarly, the inset of Figure 3.5 shows
the scatter obtained when R^ is plotted against ozone dose, whereas
plotting R^t against DOC-normaksed O3 dose seems to lead to a
continuous power relationship (Figure 3.5).
— 4E 06
0 .
t; 2E 06
°*> -
rf> X °
50 1 0
0
0
O
Ozone dose [|jM]
X
01 02 03 04 05
Ozone dose / DOC [M/M]
Figure 3.5. Ozonation at pH 8 and various ozone doses of diluted
Berlin wastewater (solid triangle), Berlin wastewater (solid squares),
Opfikon wastewater (open circles), Lake Zurich water (star)Rct after 6 seconds as a function of ozone dose normalized to DOC
concentration Inset same data with ozone dose not normalized
On one hand, the above observations are puzzling as one would predict
the nature of DOM in Berlin wastewater to be different than that of
Opfikon wastewater and even more than that of Lake Zurich water
66 Chapter 3
—i.e. that the decomposition of ozone in those waters must be related
to one another by a more complex relationship than their mere DOC
values. On the other hand, biological processes are a very important
step in wastewater treatment and might result in somewhat similar
ozone-reactive DOM fractions. Although the authors warrant caution
in making generakzations out of the small number of waters
investigated here, DOC concentration undeniably stands out as an
important empmcal parameter for the charactenzation of ozone
decomposition in wastewater.
3.4.2.2 Role of HO'scavengers
Ozone decomposition was studied in waters spiked with vanous HO'
scavengers (inhibitors and promoters) to investigate the importance of
the radical chain reaction. In Figure 3.6a, Opfikon and Berlin
wastewater (open tnangles and circles) were spiked with 1-propanol
(inhibitor) and compared to the same experiments without 1-propanol
addition (sokd tnangles and circles). The addition of an inhibitor did
not have a ma|or impact on ozone decomposition in either wastewater
suggesting that it is not controlled by a radical chain reaction. A similar
experiment was performed with a batch reactor with Lake Zunch water
spiked with tert-butanol (see squares and inset). Here the stabikzation
effect of the scavenger is only noticeable for t > 20 seconds, i.e. dunng
the second phase of ozone decomposition in natural water. It should
be noted that the DOC concentration in Lake Zunch water is small
(1.4mgC/L), so that the importance of the radical chain reaction can
be expected to be less significant than for natural waters with higher
DOC concentrations. In Figure 3.6b, the HO'-probe pCBA was
measured and confirmed that for experiments spiked with either
1-propanol or tert-butanol, all HO' had been scavenged.
Experiments were also performed with Opfikon wastewater spiked
with 60 uM methanol (promoter) which resulted in an increase in the
Characterization of ozone decomposition in wastewater 67
rate of ozone decomposition (data not shown). This indicates that if
the proportion (in the DOM) of promoters versus compounds reacting
directly with ozone is high enough, the autocatalytic chain reaction can
affect the ozone decomposition (e.g. industnal wastewater).
Nevertheless, for DOM compositions found in Berlin and Opfikon
wastewaters, the direct reactions with ozone and not the autocatalytic
chain reaction seemed to control its decomposition.
Experiments with an addition of 1 uM H2O2 (initiator) to Berlin
wastewater did not impact ozone decomposition (data not shown).
Given that only the deprotonated form HO2 reacts with ozone, the
reaction is to slow to be observed during the first 20 seconds
(pKaH202/H02 = 11.6 and k 03/H02= 2.8 x 106 M h \ then ti/2 = 980 s).
Even when the H2O2 concentration was increased to 10 uM (ti/2 = 98 s),
the increase of ozone decomposition was not statistically significant
dunng the first 6 seconds (from 93% to 96%), while pCBA oxidation
increased slightly from 33% to 40%. In a separate investigation (29),
12.5 mg/L ozone was added to Berlin wastewater containing
9 mg/L DOC and a skght increase in ozone decomposition was
observed already with 3.6 uM H2O2. In that case, however, O3 to DOC
ratio was high, allowing ozone to be measured well into the second
phase of ozone decomposition (up to 600 s), making the effect of
H2O2 addition noticeable.
The use of O3/H2O2 as an AOP for the treatment of wastewaters is
therefore only sensible at (1) high H2O2 concentrations where the
mechanism of H2O2 as an initiator becomes significant compared to
ozone's direct reactions with the DOM and/or (11) high O3 dose to
DOC ratios where O3 residual last long enough for the H202-mduced
mechanisms to become significant. It is important to recall that given
the R^t values shown earlier (see pre-03 dose section), wastewater
ozonation is already intrinsically an AOP.
Chapter 3
Ozone Exposure [Ms]1E-04
cgBaaagA
V
-& —ts—— A A
b)
*A*A
AA
A
Ozone Exposure [Ms]
0 01 0 02
^
X
i
1 5
Figure 3.6. Effect of HO" scavengers (inhibitors) on ozone
decomposition. At same experimental conditions (pH 8, 50 uM ozone),lake Zurich water w/ and w/o 12 mM tert-butanol (squares), Opfikonwastewater w/ and w/o 12 mM 1-propanol (triangles), Berlin wastewater
w/ and w/o 12 mM 1-propanol (circles). Open symbols w/ scavenger,
soHd symbols w/o scavenger, (a) ozone concentration as a function of
time. Inset: time scale increased 20 times for lake water experiments
(b) concentration of the HO"-probe pCBA as a function of ozone exposure.
Inset exposure scales increased 100 times for lake water experiments.
Characterization of ozone decomposition in wastewater 69
3.4.2.3 Effect of pH
Dunng the second phase of ozonation in natural waters, it is well
known that an increase in pH favours the autocatalytic chain
reaction accelerating ozone decomposition (i.e. increasing k).
The effect of pH variation on ozone decomposition in wastewater,
however, is not known.
Figures 3.7a,b show the effect of pH vanation on ozone
decomposition and HO' formation in Opfikon wastewater. At pH 2,
there is a slow ozone decomposition up to 10 seconds and the ozone
concentration seems to stabikze after 10 seconds. As pH increases from
pH 2 to pH 7.9, ozone decomposition rate increases rapidly
Figures 3.7b shows that at pH 2, HO' are generated pnor to
350 milliseconds but barely any pCBA decrease was measured after
that. The HO' generation increases significantly with raising pH,
however, it seems to reach a maximum at pH 6.7. An increase to
pH 7.9 did not significantly change HO' generation.
Consistently with the rest of the investigation, R<.t could be well
modelled with power functions with similar ß values (see pre-ozonation
section) at all pH values (data not shown). The lines were shifted on the
Rct axis with Rct values after 1 second (= a) of 2.2 x 10 7 at pH 2.0,
1.2 x 10 6at pH 4.1, and 2.1 x 10 6
at pH 6.7 and pH 7.9.
Confirming results of earker sections, ozone autocatalytic
decomposition seems not to play a key role during the initial phase.
Although the concentration of initiator HO increases at a higher pH,
its rate of reaction with ozone is too small to explain the observed
increased ozone decomposition at an increased pH. The effect of pH
on ozone decomposition in wastewater might therefore be attnbuted to
protonation/deprotonation of reactive species in the organic matter.
70 Chapter 3
Time [s]
10 15 20 25
S-2-
!<:'"* pH2 0
~ -A
pH4 1
N, pH 6 7
"
-o
a)^
\ pHT9
Ozone Exposure [Ms]
2 5E-04 5 OE-04 7 5E-04
pH2 0
•
l\m
*>
">.
'\ pH6 7
o b)
pH79
Figure 3.7. Ozonation of Opfikon wastewater at a dose of 2.5 mg/L(52 uM) and at pH 2.0, 4.1, 6.7 and 7.9. (a) Ozone decompositionkinetics, (b) Oxidation of the HO"-piobe pCBA as a function of
ozone exposure.
Characterization of ozone decomposition in wastewater 71
3.4.3 Impact of Ozonation on Water Quality Parameters
All expenments discussed above relate to the effect of operating and
water quakty parameters on the oxidants dynamics —O3 and HO'.
It is also of interest to gam some information on the changes of
parameters representative of DOM or of subgroups thereof.
3.4.3.1 Absorption changes In the water
It is well known that ozonation is an effective process for colour
removal. Waters with high DOC typically have a yellow colour and turn
bluer when ozonated. This can be expected as ozone reacts readily with
compounds in the DOM that absorb in the blue range of the visible
spectrum (e.g. bikrubin contnbutes to urine's colour and contains
amino groups, smax_450mn = 55'000 M ^m 1, e 285nm= 8'300 M 1cca 1).
In Figure 3.8a, the absorption of ozonated water relative to raw water
at 285 nm is plotted as a function of time (open tnangles). The largest
decrease in absorption (20%) occurs during the first 350 milkseconds,
resembkng closely ozone decomposition (solid circles). In Figure 3.8b,
ln(A/Ao) is plotted as a function of ozone exposure and a linear
relationship is obtained with k"app = 4000 M h 1. Such lineanty is
surpnsing because it would suggest that the chromophonc group or
class responsible for the absorption change at 285 nm dunng the
observed time interval has a uniform 03-reactivity (constant k").
Moreover, when measunng absorbance of DOM fractions
(with LC-OCD-OND), UV absorbance decreased in a similar fashion
for all fractions, indicating that it is unkkely that one class of organic
compounds be responsible for the decolounsation quantified above.
This linear relationship between absorption and ozone exposure could
be of practical interest as it might allow the extraction of ozone
exposure in a real system based on the empmcal quantification of
ozone induced decolounsation.
72 Chapter 3
Ozone Exposure [Ms]5E05
b)
ÖQ
y = - 4080x - 0 2
R2 =0 996
U
25 50
Figure 3.8. Changes in absorbance of Opfikon wastewater at pH 8
following 2 1 mg/L (44 uM) ozone addition (a) Absorbance of ozonated
water normalized by raw water absorbance at 285 nm and ozone
concentration versus time (b) Same data, plotted as natural logarithm of
absorbance versus ozone exposure Inset Change in water absorbance
(285 nm) versus molar ozone consumption at each time step
Characterization of ozone decomposition in wastewater 73
The spectrum of absorbance change at vanous reaction times of
ozonation was also measured. A local maximum was observed between
255 nm and 285 nm (~264 nm), which is a range in which aromatics
absorb strongly (e.g. ephenokte=1800 M^m1 at 270 nm). The inset of
Figure 3.8b shows the absorbance change (Ao-A) at 285 nm plotted as a
function of ozone consumption ([Oj]o-[03]). Given the linear
relationship, the apparent molar absorption coefficient of the
chomophonc group/class can be calculated as eapp > 1000 M 'cm 1
(e = AA/(AC X) with A = 1 cm; for eapp the sign > is used because
the stoichiometnc factor is unknown but likely > 1.0).
3.4.4 Exploratory Model
3.4.4.1 Concept
Based on the above findings, ozone decomposition in wastewater
seems not be controlled by the radical chain reaction. Therefore, we
hypothesise that it is controlled by direct reactions between ozone
and some highly reactive moieties of the dissolved organic matter.
Owing to the complex nature of the wastewater matrix and DOM
in general (22,30,31), one could expect its reactivity towards ozone
to be best described by some continuous distnbution of rate
constants and concentrations.
Ozone is known as a selective oxidant, as exemplified in Figure 3.9,
second-order rate constants (dark coloured bars) spread over more
than 10 orders of magnitude (25). The distribution does not follow
any simple law; moreover it strongly vanes with pH changes as
deprotonated species react much faster with ozone. In contrast,
HO' rate constants (light coloured bars) vary over only four orders
of magnitude with most second-order rate constants being on the
order of 109 M h 1.
74 Chapter 3
Logio (Rate Constant) [IVT's ']
Figure 3.9. Distributions of second-order rate constants for O3 (dark
grey) and HO" (light grey) obtained from (25). In the background,a model of a humic acid (HA) molecule (adapted from (22)) with the
attribution of rate constants of either oxidants (O3: black lines, HO":
grey lines) to some example moieties of the HA.
To explore the implication of various rate constants and concentration
distributions on ozone decomposition, a kinetic solver (24) was first
used to model three conceptual cases.
Figure 3.10a shows the effect of a discontinuous rate constant
distribution (107, 105, 103, 101 M 's ') on ozone decomposition on a
semi-log plot. A staircase pattern is noticeable, each step representing a
new "kinetic domain". Ozone concentration decreases slowly until it
reaches a time window where a reaction with a class of moieties is
possible at which point it decreases rapidly. When the class of moieties
has been fully oxidized, ozone concentration stabikzes until the next
time window for the next class of moieties, and so on.
Characterization of ozone decomposition in wastewater 75
Time [s]
Figure 3.10. Effect of various distributions of reactive moieties on ozone
decomposition (ozone dose = 30 uM). (a) Calculation with three
concentrations per moiety (2.5 uM, 7.5 uM and 15 mM) —all
sub-stoichiometric with respect to ozone dose— and two orders of magnitude
gaps between successive values of k" (107, 105, 103, 101 M1s1). Rising curves
represent the corresponding product formations. Vertical dashed lines show
time windows covered by CQFS and batch reactor experiments, (b) Similar to
(a) but with a finer distribution (1 moiety per order of magnitude of k") and a
concentration of 3 uM per moiety Inset data plotted on a linear time-scale.
76 Chapter 3
Figure 3.10a shows the decomposition of 30 uM ozone for three
different concentrations of moieties (2.5, 7.5, 15 uM), oxidation
products are also shown to portray the relationship between the
staircase pattern and oxidation of the individual moieties. Clearly, a
spread m rate constants and concentrations over orders of
magnitude induces a spread of kinetic occurrences over orders of
magnitude of time. In other words, it is not possible to resolve the
entire kinetic history of ozone decomposition on a linear plot
(compare with mset m Figure 3.10b).
In Figure 3.10b, a finer distnbution has been used with one rate constant
per order of magnitude between 107 and 10' M 's ' and a concentration
for each moiety of 3 uM. In this case, it is not possible to distinguish the
time windows during which moieties react because the reactions overlap
—as shown by the oxidation product pattern— and induce continuous
ozone consumption over the log of time. The inset of Figure 3.10b
exempkfies the ozone decomposition on a linear time-scale.
This representation gives the impression that some ozone "disappears"
immediately upon addition to water, hence the term "instantaneous
ozone demand" of broad use in dnnktng water ozonation.
In Figures 3.10a,b dashed vertical lines indicate the time windows that
can be investigated with the continuous quench flow system used in
this study and standard bench scale batch reactors. Although CQFS
would allow ozone concentration measurement down to
110 milliseconds (6), it is still 100 to 1000 times too slow to time-
resolve the beginning of ozone's direct reactions with the fastest
moieties (with k" ~ 107 M's' for phenolic or ammo groups at
naturally occurnng environmental concentrations). This explains why
even with CQFS, measured ozone kinetics still exhibit some "ozone
demand" prior to 350 milliseconds.
Characterization of ozone decomposition in wastewater 77
3.4.4.2 Fitting Ozone Decomposition
Figure 3.11a shows the measured decrease of ozone concentration in
Opfikon wastewater (pH 8 and [Oj]o = 83 uM (4 mg/L)) as a
function of time. Circles are data points obtained with CQFS and
squares with a batch system. Roughly 80% of the ozone is consumed
pnor to the first measurement with the batch system (vertical dashed
line). Data were then fitted with rate constants and concentration
distributions ([O3]o = 83 uM, k", = {106; 105; 104; 103; 200} M 's ',
G = {23; 10; 25; 10; 70} uM) —see black curve. Figure 3.11b shows
the same data on a semi-log plot. Based on this model, ozone
decomposition m Opfikon wastewater starts a few milliseconds after
ozone addition. The bottom curves show the succession of modelled
products formation as a function of time.
78 Chapter 3
i,^^^
b)
60-
•\
30-
v-
^ y
/'y / \n-—-^-"^'"/ \
>T
0 01 1 100
Time [s]
Figure 3.11. Ozone concentration in Opfikon wastewater at pH 8 and
4 mg/L ozone dose. Solid circles are from CFQS and sofid squares
from batch experiments. Experimental results are fitted with adequatedistributions of reactive moieties concentrations and rate constants
(black line) (a) O3 as a function of time on a linear time-scale
(b) Same as a) on a log time-scale, the sofid grey lines represent the
products formed during the "virtual experiment".
Characterization of ozone decomposition in wastewater 79
3.4.4.3 Predicted Effect of Dose on Ozone Decomposition
Given the reasonable fit obtained with the model, it is of interest to see
if the model can be used to predict ozone decomposition kinetics
obtained for specific expenmental conditions. In Figure 3.1a, data
corresponding to the expenment with [Oj]o = 40 uM was first fitted
with the model descnbed above. [Oj]o was then modified in the model
to match the doses used in two independent expenments (52 uM and
31 uM) while keeping the same distnbution of moieties. The model
predictions (sokd lines) are very close to expenmental results (symbols)
at both 52 uM and 31 uM. The ability of the model to track changes
following dose vanation indicates that changes in the observed ozone
decomposition kinetics at vanous doses —as observed m 3.4.1.1— can
be entirely explained by second-order kinetic effects.
complete ozone ,
consumption s
fcefore350m&'
40
Ozone dose [^jM]
Figure 3.12. Ozone consumption prior to 350 milliseconds versus
ozone dose in Opfikon wastewater at pH 8 (open triangles) The solid
line is based on model predictions, the diagonal represents the limit at
which all added ozone is consumed prior to 350 milliseconds
80 Chapter 3
In Figure 3.12, open tnangles show ozone consumed dunng the first
350 milkseconds after ozone addition as a function of ozone dose in
Opfikon wastewater at pH 8. The model descnbed above was utikzed
to obtain a trend kne (sokd kne) for the same conditions and agrees
well with the data. At doses < ~20 uM, all ozone is consumed by
reactive moieties dunng a time window that precedes what can be
investigated by CQFS (i.e. the curve has a slope of 1).
At doses > ~20 uM, there is enough ozone to oxidize all of the moieties
that react in less than 350 ms and the remaining ozone is consumed
dunng the time window investigated with CQFS (< 20 seconds).
In these expenments all ozone was consumed pnor to 20 seconds.
The model sensitivity to pH changes was also investigated (data not
shown). The rate constants m the amine and phenol range were
decreased by one order of magnitude for every decreasing pH unit,
starting at pH 8 for which the model had previously been fitted. There
was a reasonable agreement at lower and higher pH values (pH 2,
pH 6.7 and pH 7.9), however the model failed m the intermediate range
(pH 4 and pH 6). Nevertheless, the model agreement with the trend
adds some weight to the hypothesis that the impact of pH on
wastewater ozonation is due to the protonation of reactive species.
3.5 Conclusion
For the wastewaters investigated here, the kinetics of ozone decomposition
is not controlled by the autocatalytic chain reaction. However, if an
initiator/promoter concentration is significantly increased, the autocatalytic
chain reaction can become important. It is suggested that ozone
decomposition in wastewater is controlled by direct successive reactions of
ozone with speafic moieties of the organic matter. It is also suggested that
HO' are directly generated during those reactions which explains the high
initial concentrations of HO' measured (17).
Characterization of ozone decomposition in wastewater 81
As demonstrated, wastewater ozonation is already mtnnsically an
advanced oxidation process with very high initial HO' concentrations.
Transient HO' concentrations calculated here (0.14 nM) are more than
100 times larger than those occurnng in natural water AOP processes
such as O3/H2O2. Hence, the addition of H2O2 to wastewater does not
necessanly lead to an acceleration of O3 decomposition or an increase
in HO' generation.
Although wastewater ozonation seems not be controlled by the
autocatalytic chain reaction, it is very sensitive to pH. It is suggested
that this is due to deprotonation of reactive moieties of the organic
matter and can be reproduced by a kinetic model taking dissociation
into account.
In the studied wastewaters, absorbance decrease at 285 nm is first order
with ozone concentration. This first order kinetic relationship between
ozone and absorbance at 285 nm could have practical impkcation as, if
confirmed, the decrease in absorbance could easily be used to
back-calculate ozone exposure in real systems.
Throughout this parametnc investigation, DOC stands out as a key
water quality parameter that might be used to normalize ozone doses,
allowing for a companson of waters of vanous ongms
The charactenstics of the mechanisms involved dunng wastewater
ozonation match those of the initial phase during natural water
ozonation. It is therefore suggested that wastewater and natural water
ozonation involve the same fundamental mechanisms, the only
difference being that [Oj]o/DOC ratios are typically significantly
smaller in wastewater, preventing observation of the second phase.
82 Chapter 3
A kinetics model based on parallel bimolecular reactions of reactive
moieties with ozone was able to fit the measured ozone decomposition
kinetics. Based on the model fit, the first reactions with ozone must
start occurnng in the lower millisecond range. The effects of dose
vanation on ozone decomposition could be reproduced by the model,
which indicates that these are second order kinetics effects. Clearly, the
use of the model is purely exploratory, the complexity of dissolved
organic matter, precludes any accurate predictive modekng.
It is however a valuable tool to help support/disprove hypotheses and
improve our grasp of the complex non-linear interactions involved
dunng the numerous parallel second-order oxidation reactions
tnggered by ozone addition to wastewater.
3.6 Acknowledgments
We thank CIRSEE - Suez Environnement for financial support;
Isabelle Baudin, Auguste Bruchet, Zdravka Do-Quang, Mane-Laure
Janex, Jean-Michel Lamé and Philippe Savoye (CIRSEE) for fruitful
discussions; Michael Dodd, Marc Huber, Max Maurer, Gretchen
Onstad and Eksabeth Salhi for insightful comments. Special thanks to
Adnano Joss for his insight on the treatment processes involved at the
Opfikon WWTP and to Patnce Goosse and Sebastian Zabczynski for
their help in obtaining the water samples. Moreover, we thank the
German Ministry of Education and Research (BMBF) for supporting
the research stay of Jochen Schumacher at eawag.
3.7 References
1 Robson, C M , Rice, R G, Wastewater ozonation in the USA- historyand current status - 1989 O^one: Science & Engineering 1991, 13, 23-40
2 Ternes, T A,Occurence of drugs in german sewage treatment plants and
rivers Wat. Res. 1998, 32, 3245-3260
Characterization of ozone decomposition in wastewater 83
3 Huber, M M, Canonica, S
, Park, G -Y, von Gunten, U, Oxidation of
pharmaceuticals during ozonation and advanced oxidation processes
Environ. Sa. Technol. 2003, 37, 1016-1024
4 Huber, M M, Ternes, T A
,von Gunten, U, Removal of Estrogenic
Activity and Formation of Oxidation Products during Ozonation of 17a-
Ethmylestradiol Environ. Sa. Technol. 2004, 38, 5177-5186
5 Dodd, M C, Buffle, M-O, von Gunten, U, Moiety-specific oxidation of
antibacterial molecules by aqueous ozone Reaction kinetics and relevance
to ozone-based wastewater treatment Environ. Sa. Technol., in press, 2006.
6 Buffle, M -O, Schumacher, J, Salin, E , Jekel, M ,von Gunten, U,
Measurement of the Initial Phase of Ozone Decomposition in Water and
Wastewater by Means of a Continuous Quench Flow System Applicationto Disinfection and Pharmaceutical Oxidation Wat. Res., accepted, 2006.
7 Rakness, K L, Corsaro, K M , Hale, G, Blank, B D, Watewater
disinfection with ozone - Process control and operating results O^one Sa.
Eng. 1993, 15, 497-514
8 Paraskeva, P, Graham, N J D, Ozonation of municipal wastewater
effluents Wat. Envir. Res. 2002, 74, 569-581
9 Xu, P, Janex, M -L, Savoye, P, Cockx, A , Lazarova, V, Wastewater
disinfection by ozone main parameters for process design Wat. Res.
2002, 36, 1043-1055
10 Lazarova, V, Savoye, P, Janex, M -L, III, ERB, Pommepuy, M ,
Advanced wastewater disinfection technologies state of the art and
perspectives Wat. Sa. Tech. 1999, 40, 203-213
11 Savoye, P, Janex, M -L, Lazarova, V, Wastewater disinfection by low-
pressure UV and ozone a design approach based on water quality Wat.
Sa. Tech. 2001, 43, 163-171
12 Janex, M -L, Savoye, P, Roustan, M , Do-Quang, Z , Laine, J -M ,
Lazarova, V, Wastewater disinfection by ozone influence of water qualityand kinetics modeling O^one Sa. Eng. 2000, 22, 113-121
13 Huber, M M, Goebel, A , Joss, A , Hermann, N , Loeffler, D, McArdell,
C S, Ried, A , Siegrist, H , Ternes, T A
,Von Gunten, U, Oxidation of
Pharmaceuticals during Ozonation of Municipal Wastewater Effluents A
Pilot Study Environmental Science and Technology 2005, 39, 4290-4299
14 Hoigné, J In The Handbook of Environmental Chemistry, Hrubec, J, Ed,
Springer Verlag, 1998, Vol 5, pp 83-141
15 von Gunten, U, Ozonation of drinking water Parti Oxidation kinetics
and product formation Wat. Res. 2003, 37, 1443-1467
16 Elovitz, M S,von Gunten, U, Hydroxyl radical/ozone ratios during
ozonation processes I The Rct concept O^one Sa. Eng. 1999, 21, 239-260
84 Chapter 3
17 Buffle, M -O,von Gunten, U, Phenol and Amine-lnduced HO"
Génération During the Initial Phase of Natural Water Ozonation
Environ. Sa. Technol., accepted, 2006.
18 Buffle, M -O, Galli, S
,von Gunten, U, Enhanced Bromate Control
during Ozonation The Chlorine-Ammonia Process Environ. Sa. Technol.
2004, 38, 5187-5195
19 Joss, A , Andersen, H , Ternes, T, Richie, P R , Siegrist, H ,Removal of
Estrogens in Municipal Wastewater Treatment under Aerobic and
Anaerobic Conditions Consequences for Plant Optimization Environ. Sa.
Technol. 2004, 38, 3047-3055
20 Schumacher, J , Pi, Y Z, Jekel, M ,
Ozonation of persistent DOC in
municipal WWTP effluent for groundwater recharge Water Sa. Technol.
2004,42,305-310
21 Bader, H , Hoigné, J ,Determination of ozone in water by the indigo
method Wat. Res. 1981, 15, 449-456
22 Buffle, J Complexation reactions in aquatic systems: an analytical approach, Ellis
Horwood Limited Chichester, 1988
23 Huber, S A, Frimmel, F H ,
A New Method for the Characterization of
Organic-Carbon in Aquatic Systems InternationalJournal of Environmental
Analytical Chemistry 1992, 49, 49-57
24 Braun, W, Herron, J T, Kahanar, D K, ACUCHEM Computer programfor modeling complex réaction systems Int. J. Chem. Krnet. 1988, 20, 51-62
25 NIST, http //kinetics mst gov//solution/index php A compilation ofkinetics data on solution-phase reactions 2002
26 Rosenfeldt, E J , Linden, K G, Dégradation of Endocrine DisruptingChemicals Bisphenol A, Ethinyl Estradiol, and Estradiol during UV
Photolysis and Advanced Oxidation Processes Environmental Saence and
Technology 2004, 38, 5476-5483
27 Acero, J L, von Gunten, U, Characterization of oxidation processes
ozonation and the AOP 03/H202 Jour. AWWA 2001, 93, 99-100
28 Elovitz, M S,von Gunten, U, Kaiser, H -P, Hydroxyl radical/ozone
ratios during ozonation processes II The effect of temperature, pH,
alkalinity and DOM properties O^one Sa. Eng. 2000, 22, 123-150
29 Schumacher, J , Stoffregen, A , Pi, Y, Jekel, M ,Use of the Rct concept
for the description of the oxidation potential of ozone towards effluents
of wastewater treatment plants Vom Wasser 2004, 102, 16-21
30 Buffle, J, Huang, P M, Senesi, N Structure and surface reactions of soil
particles, John Wiley & Sons, 1998, Vol 4
31 Dignac, M -F, Caractensation chimique de la matière organique au cours
du traitement des eaux usées par boues activées These de Doctorat de
l'Université Paris L71998
4 Phenols and Amines Induce HO'
Generation During the Initial Phase
of Natural Water Ozonation
Marc-Okvier Buffle and Urs von Gunten,
EnvironmentalScience and Technology, 2006
4.1 Abstract
The initial phase of ozpne decomposition in natural water (t<20s) is poorly
understood. It has recently been shown to result in very high transient HO'
concentrations and thereby plays an essential role during processes such as bromate
formation or contaminants oxidation. Phenols and amines are ubiquitous moieties of
natural organic matter. Naturally occurring concentrations of primary, secondary and
tertiary amines, amino aads and phenol were added to surface water and ozpne
decomposition as well as HO'generation were measured starting 350 milliseconds after
ozpne addition. Six seconds into theprocess, 5 jjlM of dimethylamine andphenol had
generated IHO'dt = 1x1Ow Ms and 1.8x10'° Ms respectively With 10 /uM
àmethylamine and 1.5 mgOJL, Ra (]HO'dt/\03dt) reached 10e: larger than in
advanced oxidation processes (AOP) such as 0JH202. Experiments in presence of
HO'-scavengers indicated that a significant fraction of phenol-induced ozpne
decomposition and HO' generation results from a direct electron transfer to ozpne.
For dimethylamine, the main mechanism of HO' generation is àrect formation of
0/ which reacts selectively with 03 to form 0'. Pretreatment of phenol-containing
water with HOCl or HOBr did not decrease HO' generation, while the same
treatment of dimethylamine-contaimng water considerably reduced HO'generation.
4.2 Introduction
Ozone is a strong oxidant used in water treatment for taste, odour and colour
removal, oxidation and disinfection. Current research also demonstrates that
emerging contaminants such as cyanotoxins, hormones and antibiotics are
efficiently oxidized and biochemically inactivated by ozone (1-5).
86 Chapter 4
4.2.1 Biphasic kinetics of ozone decomposition.
Ozone decomposition in natural water can be kinetically and
mechanistically divided into an initial and a second phase (6-8)
(Figure 4.1a, adapted from (7J).
Dunng the initialphase (t < ~20 s), rapid direct reactions of ozone with
specific NOM moieties and some inorganic compounds consume a large
fraction of the added ozone (often called instantaneous ozone demand).
Dunng this phase, ozone does not follow an apparent first-order rate law as
during the second phase (Figure 4.1a). In fact, k'o3 [s x\ increases following
a power function with t —» 0 (6). Very high yields of HO' are generated and
Rct (=jHO'dt/J03dt) also increases following a power function
with t —> 0 (7). Interestingly, although HO' transient concentrations are
very high during the initial phase, ozone decomposition seems not to be
controlled by the radical chain reaction as in the second phase (6).
During the secondphase ozone decomposition follows an apparent first-order
rate law, i.e. k'o3 [s J remains constant and is 10 — 100 times smaller than
during the initial phase. R« is also 10 — 100 times smaller and constant (7).
The most reactive moieties of NOM have reacted with ozone during the
initial phase so that ozone decomposition in the second phase is mostly
controlled by a radical chain reaction and not by its direct reaction with
those moieties (Figure 4.1b). The radical chain reaction is descnbed
exhaustively elsewhere (9,10). In short, the reaction is partially initiated by O3
reaction with HO to form HO2. O3 reacts again with HO2 and generates
O2*, which reacts very selectively with another O3 to form O3', readily
generating HO'. HO' reaction with certain NOM moieties (promoters)
leads to carbon centered radicals which upon O2 addition form more O2*,
and so on. This part of the reaction sequence is called propagation. The
chain reaction is terminated upon the reaction of HO' with compounds
(inhibitors, e.g carbonate, t-butanol) that do not lead to O2* formation.
Phenols and amines induced HO' generation 87
Initial Phase Second Phase
NOM0,
Indirect 02" or
H202 gen.= promoters
b)
Figure 4.1. Initial and second phase of ozone decomposition in natural
water (a) Measurements with CQFS for t<20s (open circles) and batch
experiments for t>20s (solid squares) in Lake Zurich water at pH 8 and
2.4 mg03/L (50 uM). (b) During the initial phase some NOM moieties react
directly with ozone to form O2* or O3*, while during the second phasesome NOM moieties promote the radical chain induced ozone
decomposition by reacting with HO and subsequently O2 to release O2 (9).
88 Chapter 4
4.2.2 Importance of the initial phase
Given the charactenstics of the initial phase —rapid ozone
decomposition and very high relative concentration of HO'—
its investigation is essential to better comprehend a number of key
mechanisms involved in water ozonation. For example, the formation
of potentially carcinogenic bromate dunng ozonation of bromide-
contaming water is in part due to HO' generated dunng the initial
phase (11). By reducing HO' exposure dunng the initial phase through
pre-chlonnation and ammonia addition, bromate formation could be
significantly decreased (11). Another example is that of certain
compounds considered refractory to ozone oxidation (e.g. îopromide)
but nevertheless showing significant degradation dunng wastewater
ozonation (3). This effect can be well explained by very high HO'
exposures measured dunng ozone decomposition in wastewater, which
is mechanistically similar to the initial phase m natural water (6,7).
A further example is ozone's specificity —which though advantageous
when selectively oxidizing biochemically active moieties of
pharmaceutical compounds— might be compromised dunng the initial
phase if large fractions of compounds are oxidized by unselective HO',
thus inducing primary metabolites that might not be biochemically
inactive (for an extensive discussion see (5)).
In this paper we investigate mechanisms responsible for the initial
phase and propose moieties of the NOM that might simultaneously
generate high ozone decomposition rates and high HO' yields.
4.2.3 Presence of ozone-reactive moieties in NOM.
Natural organic matter (NOM) consists of an infinite vanety of
organic molecules, ranging from low (<100 g/mol) to high molecular
weight compounds (e.g. humic acid MW ~ lO'OOO g/mol). Most
functional groups known to organic chemistry are found in NOM.
Phenols and amines induced HO' generation 89
Comprehensive reviews can be found elsewhere (12,13). Phenolic
moieties in lake NOM can be estimated to be in the range of 0.5 to
10 uM depending on the proportion of pedogemc (more aromatic)
versus aquagenic (more akphatic) denved NOM (12). The sum of all
ammo acids has been measured m the range of 4 to 12 uM in low
productivity lakes (12). Concentrations of organic amine moieties are
typically not known, with the exception of hexosammes which have
been measured in the range of 0.18-0.85 uM in low productivity lakes
(12). Hexosammes, however, are typically acetylated m the natural
environment, which essentially inactivates them with regard to ozone
oxidation; ozone reacts very slowly with amides and sacchandes in
general. Dissolved organic nitrogen (DON) might give a rough
estimate of the maximum possible amine concentration; it has been
measured m the range of 0.07 uM to 35 uM for low productivity
lakes (12). Unfortunately, DON is typically obtained by subtraction of
nitrate fiom total dissolved nitrogen (TDN) resulting in rather poor
estimates given the innocuous presence of comparatively high
concentrations of nitrate in surface water. Moieties of importance to
ozone, such as olefins, are not reported. In summary, it can be assumed
that natural waters contain concentrations in the uM range of phenolic
and amine moieties m the NOM. It is, however, clearly not possible at
this point to provide an exact descnption of the concentration
distnbution of ozone-reactive moieties m natural water matrixes.
4.2.4 Ozone-reactive functional groups.
Ozone is a specific oxidant, which reacts readily with a kmited number
of functional groups such as olefins, amines and activated aromatic
systems (10). As discussed above some of these moieties can be found
m uM concentrations in lake NOM. Ozone reacts readily with olefins
through cyclo-addition (e.g. for non-substituted olefins: k" ~105 M h 1).
These reaction rates are nearly independent of pH and neither O2* nor
90 Chapter 4
O3' is generated in the process (14). The apparent reaction rates of
phenols and amines with ozone display very strong pH
dependencies. Protonated species react many orders of magnitude
slower than the deprotonated (phenolate) or neutral species (amine).
Recent investigations have shown that amines and phenols generate
HO' through formation of O3' which at neutral pH instantaneously
decays into HO' and O2. An indirect pathway leads to the
generation of O2* which reacts quickly and selectively with O3 to
form 03' and finally HO' (15-17).
Tertiary amines generate ~10% O3' upon an electron transfer to ozone
following eq 1. However, the mam mechanism leads to the formation
of oxylamme and oxygen following eq 2 (15).
R3N + O3 -> R3N'+ + 03' (1)
R3N + O3 -> R3NO + !02 (2)
In contrast, the mam mechanism of the reaction between ozone and
secondary amines, seems to induce O2* (eq 3), while only 20% leads to
hydroxylamme and singlet oxygen (eq 4) (18).
R2NH + O3 -> R2NO' + H+ + 02' (3)
R2NH + O3 -> R2NOH + 1O2 (4)
In the case of pnmary amines and ammo acid, the mechanisms are less
well understood, but are probably akin to the secondary amine
mechanism. Based on the above mechanisms, one can expect the
reaction of ozone with tertiary amines to generate HO' more rapidly
but m smaller yields than in the reaction with secondary amines.
Oxidation of secondary amines should also consume significantly more
Phenols and amines induced HO' generation 91
ozone than that of tertiary amines due to the O2* intermediate which
reacts with an additional ozone molecule pnor to generating HO'.
A fraction of the reaction of phenol (i.e. phenolate at neutral pH) with
ozone also generates HO' following an electron transfer forming O3'
with a yield of 22%, as mdicated by eq 5 (17). Subsequent oxidation of
its product with the generated HO' can lead to formation of O2*,
accelerating ozone decomposition.
PhO + O3 -> PhO' + 03' (5)
In this investigation, naturally occurnng concentrations of pnmary,
secondary and tertiary amines, ammo acids and phenol were added to
surface water and ozone decomposition, as well as HO' generation
were measured starting 350 milliseconds after ozone addition.
4.3 Materials and Methods
4.3.1 Reagents.
All reagents were of analytical grade. All solutions were made from
milhQ water with a resistivity > 18 mü Ozone stock solution was
made by sparging an ozone-oxygen gas mixture through ice-cooled
water. The solution was diluted to reach O3 = 500 uM, acidified
with 1 mM H2SO4 and kept at 1°C in an ice bath for the duration
of the experiments. Indigo reagent was 312 uM and
10 mL/L H3PO4 (85%) for all expenments. Fulvic and humic acid
isolates were obtained from the International Humic Substance
Society (IHSS, Nordic aquatic FA and HA).
92 Chapter 4
4.3.2 Natural waters
Lake Zunch water (pH 7.8, alkalinity 2.4 mM, DOC 1.4 mg/L) was
collected from the raw water intake of Zunch drmkmg water treatment
plant, 30 meters below the lake's surface. The water was filtered at
0.45 urn and kept at 4°C. Waters were buffered with 0.5 mM borate
which increased the pH to 8.6. After mixing with ozone solution (1:10)
the final pH was 7.95 + 0.05. Para-chlorobenzoic acid (pCBA, 1 uM)
was added to all waters to serve as a HO' probe.
4.3.3 Methods
HO' exposure was back-calculated based on the extent of oxidation of
a pCBA, analyzed with HPLC (8). Ozone was measured onkne with a
Vanan Cary 100, either directly at 258 nm (e = 3000 M 1aca a) or with
indigo at 600 nm (e = 20'000 M'cm1) (19). When HO' induced
reactions needed to be excluded, 50 mM tert-butanol (tBA) was added
to the solution to serve as an HO' scavenger (>99% HO'-scavengmg,
controlled with pCBA). The rapid measurement of ozone
decomposition was performed with a continuous quench flow
system (CQFS). The system which has been descnbed and
charactenzed previously (6,7) rapidly mixes a stream of aqueous ozone
with one of natural water m a contact loop and quenches residual
ozone with an mdigo reagent. CQFS allows a first measurement
115 milliseconds after ozone addition. For this pro|ect the system was
slightly modified with the mdigo double-syringe pump being replaced
by a large volume (0.266 L) smgle-synnge pump (ISCO260D) which
reduced noise induced by synnges switchover. Measurements with
CQFS display the following statistics: 90% confidence intervals of
ozone concentration measurements m ozone decomposition kinetic
expenments are on average 11% off the mean values and pCBA displays
90% confidence intervals on average 5% off the mean values (7).
Phenols and amines induced HO' generation 93
4.4 Results and Discussion
4.4.1 Effect of phenolic and amino compounds on HO*
generation and ozone decomposition
In Figure 4.2a, ozone decomposition in Lake Zunch water at pH 8
(solid triangles) shows that 10% of ozone is consumed prior to
350 milliseconds and 40% prior to 20 seconds. This is in agreement
with earlier research (7). In a standard ozone batch experiment with
Lake Zunch water at pH 8, the so-called "instantaneous ozone
demand" (IOD) would therefore be calculated as 40% of the ozone
dose. The curved line displayed by the data shows that unlike what
is typically observed during the second phase, ozone kinetics in the
initial phase is not of apparent first order. Figure 4.2b shows the
increase of HO' exposure as a function of O3 exposure (JHO'dtversus JOjdt); the slopes of the curves represent R^t.
Phenol addition to Lake Zurich water (2.5 uM, crosses in
Figure 4.2) increases the ozone decomposition rate prior to the first
measurement at 350 milliseconds. Accordingly, given
kpho3 = 1.8xl07M1s1 at pH 8 (20), phenol must be completely
oxidized well before the first measurement at 350 milliseconds
(i.e. 99% at 8 milliseconds). Following this rapid reaction, however,
the rate of ozone decomposition is similar to that in Lake Zunch
water. This indicates that two orders of magnitude of time after
phenol oxidation is completed, the radical chain reaction is not
substantially accelerated. The generation of HO' in Figure 4.2b
(crosses) concurs with this description; a very high generation of
HO' occurs prior to 350 milliseconds followed by a rather sluggish
one, similar to that of Lake Zurich water.
94 Chapter 4
10Time[s]15
u -
~
"""""———w___m____Zunch water
--—-—__a glycine
1 -
^—---——__.___w phenol
'
—-~—^a^imethylamine
glucosamine
2-
a)
\ dimethyiamine
dimethylamine
R2=0 992
I/)
"2E-10 -
T3
ÖI
0 / sé^'^ JH glucosamine
V)
OQ.
<D 1E-10- o
-/"^ /g? R2=0 997
q/ a glycine
bX
"^^^b)
OE+00-
2E-04 4E-04
03 exposure j03dt [Ms]
Figure 4.2. (a) Ozone decomposition and (b) HO" exposures duringozonation of Lake Zurich water at pH 8 and 1 5 mg03,/L (31 uM)With addition of 5 uM glycine, 2 5 uM phenol, 10 uM trimethylamine,10 uM glucosamine, 10 uM dimethylamine Percentage values
associated with solid grey symbols represent the calculated degree of
oxidation of compounds
Phenols and amines induced HO' generation 95
Tnmethylamme (open tnangles) —the most reactive amme investigated
here (kR3N 03= 5xl04 M h 1 at pH 8 (20))— has reacted substantially
pnor to 350 milliseconds (48 %) and almost completely at
3 seconds (95%). This can be observed in the ozone decomposition
profile (Figure 4.2a) for which the rate decreases considerably around
3 seconds and becomes very similar to that of Lake Zunch water
afterwards. Similarly to phenols, this indicates that tnmethylamme does
not increase the radical chain induced ozone decomposition once its
own primary oxidation is completed. Moreover, Figure 4.2b indicates
that tnmethylamme increases the HO'-exposure dunng its reaction with
ozone but not substantially once the reaction is completed (curve
becomes parallel to Lake Zunch water).
Glycine (solid diamonds) and glucosamine (open circles) are
considerably slower in their reaction with ozone. Only 70% of
glycine —kgiy 03= 1600 M h l at pH 8 (20)— is oxidized prior to
the last measurement point at 20 seconds. The rate constant for
glucosamine and ozone is not known but as a primary amine, it is
likely to be on the order of ~103 M h 1 (e.g. kbutyiNH2 03= 340 M h 1).
Ozone decomposition profiles in the glycine and glucosamine
experiments reflect these lower reactivities. Nevertheless, the extent
of ozone decomposition after 20 seconds of reaction with
glucosamine suggests that O2* is generated during the reaction. The
induced HO'-exposure also increases steadily compared to Lake
Zurich water dunng the time window of their oxidations by ozone.
Dimethylamine (solid square) —kR2NH 03= 2xl04 Mis1 at pH 8
(20)— is slightly slower to react with ozone than tnmethylamme
and is therefore oxidized principally during the time window
investigated here (25% has reacted pnor to 350 milliseconds and
99% at 20 seconds). Dimethylamine induces a very strong increase
in HO'-exposure. In companson to Lake Zunch water (26x10 u vs.
96 Chapter 4
5x10 u Ms) the enhancement is so large that the oxidation by HO' of
an ozone-refractive compound such as atrazme (kaxnamt 03= 6 M h l,
kitazine ho-= 3xl09 M h 1 (10)) would increase from 15% to 55% in
the first 20 seconds if 10 uM (0.24 mgC/L) dimethylamine were
added to Lake Zurich water. For the same expenment,
Rct (= jHO'dt/J03dt) was constant at -106 during the first
20 seconds which is high even for standard O3/H2O2 AOP and two
orders of magnitude larger than for natural water ozonation (7).
The larger HO' exposures obtained dunng ozonation of
dimethylamine concur with earlier mechanistic investigations in
synthetic water indicating an 80% O2* yield (18). Ozone
decomposition is initially more rapid with tnmethylamme but finally
more extensive with dimethylamine. This is due to the additional
reaction of ozone with O2* generated upon ozonation of
dimethylamine but not generated upon ozonation of
tnmethylamme.
Sorbic acid (10 uM, data not shown) —ksorbic 03= 9.6xl05 M h 1 at
pH 8 (4)— was also added to Lake Zunch water to model a
compound that should not impact HO' generation. Sorbic acid
should react with ozone mainly through a cyclo-addition to its double
bonds (no O3' or O2* generated) (14). As predicted, data showed
similar HO' generation after addition of 10 uM sorbic acid to Lake
Zurich water as in unmodified Lake Zunch water (data not shown).
Phenols and amines induced HO' generation 97
+ 5|iM 1 8E-10
OE+00 1E-10 2E-10
HO' exposure = jHO'dt [Ms]
Figure 4.3. HO- exposures at Jo3dt = ~12xl04 Ms followingozonation after standard addition of various compounds to Lake
Zurich water at pH 8 and 1 5 mg03/L (31 uM) Addition of 0 uM,
5 uM and 10 uM trimethylarmne, glycine, dimethylamine, and 0 uM,
2 5 uM, 5 uM phenol
Spiked concentrations of some compounds were varied to confirm
by trends the observations made above. Figure 4.3 shows the effect
on HO' exposure of 5 and 10 uM addition of tnmethylamme,
glycine and dimethylamine to Lake Zurich water. There is a clear
correlation between the compounds concentrations and HO'
exposures (at same ozone exposures ~1.2 x 104 Ms).
Dimethylamine displays the highest HO' exposure of all amines
tested. However, the above values cannot be used to directly
compare the efficiency of one molecule to generate HO' versus
another, because for the same ozone exposure, compounds will have
reacted to various degrees depending on the magnitude of their rate
constants. Spiked concentrations of phenol are also directly related
to HO' exposures generated. For the investigated ozone exposure,
the addition of 5 uM phenol increases HO' exposure 385% when
compared to unmodified Lake Zurich water.
98 Chapter 4
4.4.2 Effect of humic and fulvic acid on ozone
decomposition and HO* generation
As clearly demonstrated by the above data, certain compounds and
hence specific moieties in the NOM not only display high reactivity
with ozone but also generate high yield of HO' upon their oxidation.
It is of interest to compare the profile of HO' generated by these
compounds to those generated by fractions of NOM.
In Figure 4.4, the addition of 5 uM (60 ugC/L) fulvic acid (solid
circles) and humic acid (open circles) to Lake Zurich water shows
the importance of those NOM fractions on initial HO' generation.
Although their ozone decomposition profile cannot be
differentiated and are only marginally faster than ozone
decomposition in unmodified Lake Zurich water, the initial increase
in HO' exposures compared to Lake Zurich water is substantial.
When compared to the simple model compounds discussed above
(represented by trend lines in Figure 4.4b), it seems that the mam
effect of humic and fulvic acid on HO' generation takes place pnor
to 350 milliseconds. This demonstrates that the critical moieties in
humic and fulvic acids inducing HO' upon ozonation have rate
constants > ~ 50'000 M 1s 1. Given the known high degree of
aromaticity of fulvic and humic acid and the rapid drop in
absorption at 285 nm, during the first 350 milliseconds following
ozone addition (6), those moieties can be hypothesized to be
phenolic or/and other activated aromatics systems.
Phenols and amines induced HO' generation 99
V
o
o
• Ao
i
10Time[s]15
OZurich water
A
Ow/humic acid
b)/
/
2
/
/(CH3)2NH"2E-10 - /
"a
*o
//
/
I .--Phenol
aj /
d / -•''w / .•'oCL
xw/fulvic acid
0) 1E-10 -
-V
/ o»Ö t w/humic acid
HI // % _^-~ A
•£ Ä _T-~TCH3)3N Zurich water
aA*A
OE+00-
OE+00 1 E-04 2E-04 3E-04 4E-04 5E-04
03 exposure J03dt [Ms]
Figure 4.4. (a) Ozone decomposition and (b) HO" exposure in Lake
Zurich water at pH 8 and 1.5 mg03/L and spiked with either
5 uM fulvic acid or 5 uM humic acid.
100 Chapter 4
4.4.3 Effect of generated HO* on ozone decomposition
During an earlier investigation of ozone decomposition in
wastewater, the addition of an HO' scavenger did not modify ozone
decomposition kinetics substantially, suggesting that the radical chain
induced ozone decomposition may not have a strong influence on the
initial phase (6). In Figure 4.5, an HO' scavenger (50 mM
tert-butanol, tBA) was added to unmodified Lake Zurich water.
HO'-probe, pCBA, did not decrease during ozonation which
confirmed that all HO' reacted with tBA (data not shown). Similarly
to what has been observed earlier (6), addition of a scavenger did not
stabilize ozone dunng the first 20 seconds (compare open vs. solid
diamonds). In water spiked with 2.5 and 5 uM phenol, however, an
important reduction of the initial ozone decomposition
(< 350 milliseconds) is achieved when adding tBA (open squares and
triangles) but not in the time window between 350 milliseconds and
20 seconds. Clearly, a substantial fraction of ozone reacts directly
with phenol, undisturbed by the presence of tBA. However, the
fraction of ozone that does not react following the addition of tBA,
demonstrates that a large amount of ozone is decomposed by a
radical chain reaction. Interestingly, the temporanly higher rate of the
radical chain reaction due to the high HO' yield dunng phenol
ozonation is not sustained once phenol oxidation is complete
(i.e. ozone decomposition profiles are very similar beyond 350 ms).
The addition of tBA to the dimethylamine solution shows that up to
10 seconds, no increase in ozone decomposition can be observed due
to the radical chain reaction even though an increase in HO' exposure
can be measured well before that (solid square in Figure 4.2).
The fact that tBA decreases more the phenol- than the dimethylamme-
mduced ozone decomposition indicates that a significant mechanism in
the phenol reaction is an electron transfer mechanism. Phenol's
Phenols and amines induced HO' generation 101
electron transfer mechanism can only generate O2* via the radical chain
reaction (1 e tBA stops O2* formation), while dimethylamme's main
mechanism is a direct formation of O2*, which readily consumes an
additional O3 molecule undisturbed by the presence of tBA
(1 e tBA cannot stop O2* formation)
Figure 4.5. Ozone decomposition with and without HO" scavenger
for various compounds added to Lake Zurich water at pH 8 and
1 5 mg03/L (31 uM) Solid symbols without HO" scavenger,
corresponding open symbols addition of 50 mM tert-butanol as
HO" scavenger
4.4.4 Effect of pretreatment with HOCI or HOBr on ozone
decomposition and HO* generation
In a previous study, the importance of initially generated HO' on the
formation of bromate dunng ozonation of bromide-contammg water
was clearly demonstrated (11) Pre-chlonnation of natural water
decreased HO' and hence bromate formation significantly—even when
containing ammonia which masked chlorine as chloramme In Figure 4 6,
Lake Zunch water pre-oxidized (until complete reaction of oxidant) with
102 Chapter 4
15 uM HOC1 (open circles) or 15uM HOBr (open diamonds) is
compared to unmodified Lake Zunch water (sokd squares).
HO' exposure is plotted as a function of ozone exposure for both waters
and the HO' exposure ratio (broken kne) shows that the chlorinated water
generates only ~35% of the HO' exposure generated m unmodified
water. R« can be denved fiom the slopes of the data m Figure 4.6, and
also shows a factor of ~3 between halogenated versus non-halogenated
water. The HOBr curve is close to that of HOQ, indicating that both
halogenation processes have similar effects on 03-reactive moieties
responsible for HO' generation. One practical outcome of this finding is
that chlormation pnor to ozonation has the consequence of considerably
decreasing HO'-based oxidation processes. In some appkcations, HO' is
undesirable due to its unspecific reactivity which might lead to a more
diverse product distnbution dunng oxidation of micropollutants as well as
an increased disinfection by-product formation.
I lake Zurich water
OE+00 -
0 0 E+00
Ow/HOCI
Ow/HOBr
8s>o
2 5 E-04 5 0 E-04
03 exposure J03dt [Ms]
- 0%
E-04
Figure 4.6. Effect of natural water matrix halogenation on HO" exposure
Lake Zurich water was pre-oxidized with 15 LiM HOQ (1 mgCL/L) and
15 uM HOBr at pH 8 and subsequendy ozonated with 1 5 mgOa/LDashed line %-reduction in HO" exposure due to chlormation
Phenols and amines induced HO' generation 103
It is then of interest to investigate if halogenation of some of the
individual compounds discussed above impact the induced HO'
generation upon ozonation. The top halves of Figure 4.7 show HO'
exposure as a function of time (series order same as when plotted
against ozone exposure), and the bottom halves show ozone
decomposition as a function of time. As discussed above,
chlonnation of unmodified Lake Zunch water decreases
Rct (jHO'dt/JOadt). It can be seen when comparing solid (no HOQ)
to open square symbols (with HOQ) that the decrease in R^ is due to
both, a decrease in HO' exposure by a factor of 2 and increased
ozone stabikty (i.e. increased JOjdt).
In Figure 4.7a, for Lake Zunch water spiked with 2.5 uM phenol it is
interesting to note that the HO' exposure does not decrease after
chlormation (solid vs. open circles, on top of each other), even
though significant Q substitution can be expected to have taken
place (21,22). This is in agreement with experiments in synthetic
water that demonstrated a HO' yield of 27% upon ozonation of
pentachlorophenol versus 22% for non-substituted phenol (17).
The expenments were reproduced with 15 uM HOBr (data not
shown) with similar effects as for HOQ. The strong decrease in HO'
generated upon ozonation of pentabromophenol versus
pentachlorophenol (2% HO'-yield for BrsCöO versus
27% for QsQO) observed in (17) could not be confirmed here.
However, the halogenation with 15 uM HOX of 2.5 uM phenol is
unlikely to form such highly substituted phenols as
pentabromophenol because a significant fraction of HOX reacts with
the water matrix. A lower degree of substitution is likely to decrease
the difference between Q and Br substituent effects on HO'-yield.
104 Chapter 4
• ZH + Phenol
OZH +Phenol+ HOCI
ZH
DZH+HOCI
Time [s]
2E10 -
b) A^---^"
1E10 -
Wè^T A
too
A
D20
fiUQ^ D a D
05 - A —
A
1 5 - iZH + DMA
AZH + DMA + HOCI ^\ZH
DZH+HOCI
Figure 4.7. Effect of halogenation on HO" exposure and
O3 decomposition in Lake Zurich water spiked with dimemylamine
(DMA) or phenol at pH 8 and 1 5 mg03/L (a) 2 5 uM phenol
(b) 10 uM dimemylamine Open symbols corresponding waters
pre-halogenated with 1 mgC^/L as HOQ (15uM)
Phenols and amines induced HO' generation 105
The faster decomposition of ozone in the chlonnated water spiked
with phenols is difficult to interpret. Although halophenolates react
slower with ozone than non-halogenated phenolates, halogen
substitution leads to a pKa. depression of the phenol/phenolate pair
resulting in a higher apparent reactivity with ozone at pH 8
(e.g. kphenokte = 1.4 X 109 M h 1 While k2 chlorophenokte= 2 X 108 M h \
but p.Kph = 9.9 while pK2 ciPh= 8.3, at pH 8 this results in
kapp_ph = 1.8 x 107 M^1 while kapp_2 ciPh= 6.6 x 107 Mh1).
In other words, for lower degrees of substitution, the effect of
deprotonation on the apparent rate constant of ozone's reaction with
phenol is more important than the decrease in reactivity of the
neutral and ionic species induced by the halogen's electron
withdrawing ability. A faster reaction of ozone with chlorophenol at
pH 8 should however not change the ozone decomposition profile as
all of the phenol will have reacted long before 350 milliseconds.
The effect of halogenation on dimethylamme-spiked Lake Zunch
water is shown in Figure 4.7b. As discussed earlier, HO' exposures
following ozonation of dimethylamine-spiked water are very high
(solid triangles). Unlike for phenols however, once halogenated HO'
exposures drop roughly by a factor of 4 (open tnangles)
—experiments with 15 uM HOBr gave similar results.
Ozone decomposition is much slower in the halogenated amine
solution, indicating a significant decrease in O2* generation.
Scheme 1 shows the pathway of superoxide formation following
ozone reaction with a secondary amine. Ozone attacks the lone
electron pair at nitrogen resulting in the formation of an amme-oxyl
radical, a superoxide and a proton (18). If HOQ is added to the
solution pnor to ozonation, it reacts with the secondary amine and a
Q-substitution occurs at nitrogen (23,24).
106 Chapter 4
R R
/ /R /H,C H2Ç H,C
\ \.
N H + 03 ». CH,—fc O - N O + 02" + H
/ / I \ /H2C r' ^ b O H2C
R R
R R
/ /HnC HnC
\ \N H + HOCI ». N Cl + H20
/ /HnC HnC
\ \R R
Scheme 1. Top formation of superoxide, O2", upon reaction of a
secondary amine with ozone Bottom Cl-substitution upon reaction of
HOQ with a secondary amine hindering O3 attack at nitrogen
It is proposed that the electron withdrawing effect of chlonne
substitution at nitrogen decreases significantly the availability of
nitrogen's lone electron pair to ozone attack, hmdenng the formation
of the amme-oxyl radical and superoxide.
4.5 Acknowledgments
We thank CIRSEE - Suez Environnement for financial support;
Isabelle Baudm, Auguste Bruchet, Zdravka Do-Quang, Mane-Laure
Janex, Jean-Michel Lamé (CIRSEE) for fruitful discussions; Michael
Dodd, Gretchen Onstad and Lisa Salhi for insightful comments.
4.6 References
1 Huber, M M, Goebel, A , Joss, A , Hermann, N , Loeffler, D, McArdell,
C S, Ried, A , Siegrist, H , Ternes, T A
,Von Gunten, U, Oxidation of
Pharmaceuticals during Ozonation of Municipal Wastewater Effluents A
Pilot Study Environmental Science and Technology 2005, 39, 4290-4299
Phenols and amines induced HO' generation 107
2 Hoeger, S J, Dietrich, D R, Hitzfeld, B C,Effect of ozonation on the
removal of cyanobacterial toxins during drinking water treatment
EnvironmentalHealth Perspectives 2002, 110, 1127-1132
3 Westerhoff, P, Yoon, Y, Snyder, S, Wert, E, Fate of Endocnne-Disruptor,Pharmaceutical, and Personal Care Product Chemicals during Simulated
Drinking Water Treatment Processes Environ. Sa. Technol. 2005
4 Onstad, G D, Strauch, S, Menluoto, J, Codd, G, von Gunten, U,
Selective Oxidation of Cyanotoxins by Ozonation Treatment Environ. Sa.
Technol. submitted
5 Dodd, M C, Buffle, M-O, von Gunten, U, Moiety-specific oxidation of
antibacterial molecules by aqueous ozone Reaction kinetics and relevance to
ozone-based wastewater treatment Environ. Sa. Technol, in press, 2006.
6 Buffle, M-O, Schumacher, J, Meylan, S, Jekel, M, von Gunten, U,
Ozonation and Advanced Oxidation of Wastewater Effect of O3 Dose,
pH, DOM and HO'-scavengers on Ozone Decomposition and HO"
Generation O^one Sa. Eng, accepted, 2006.
7 Buffle, M-O, Schumacher, J, Salhi, E, Jekel, M, von Gunten, U,
Measurement of the Initial Phase of Ozone Decomposition in Water and
Wastewater by Means of a Continuous Quench Flow System Appkcaüon to
Disinfection and Pharmaceutical Oxidation Wat. Res., accepted, 2006.
8 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios dunng ozonation
processes I The Rct concept O^one Sa. Eng. 1999, 21,239-260
9 Hoigné, J In The Handbook of Environmental Chemistry, Hrubec, J, Ed,
Springer Verlag, 1998, Vol 5, pp 83-141
10 von Gunten, U, Ozonation of drinking water Part I Oxidation kinetics
and product formation Wat. Res. 2003, 37, 1443-1467
11 Buffle, M-O, Galli, S, von Gunten, U, Enhanced Bromate Control
during Ozonation The Chlorine-Ammonia Process Environ. Sa. Technol.
2004, 38, 5187-5195
12 Buffle, J Complexation reactions m aquatic systems: an analytical approach, Elks
Horwood Limited Chichester, 1988
13 Fnmmel, F H, Abbt-Braun, G, Heumann, K G, Hock, B, Luedemann, H
D, Editors Refractory Organic Substances m the Environment, 2002
14 Dowldeit, P, von Sonntag, C, Reaction of Ozone with Ethene and Its
Methyl- and Chlorine-Substituted Derivatives in Aqueous Solution
Environmental Science and Technology 1998, 32, 1112-1119
15 Munoz, F, von Sonntag, C,The reactions of ozone with tertiary amines
including the complexing agents mtrilotnacetic acid (NTA) and
ethylenediaminetetraacetic acid (EDTA) in aqueous solution Perkm 2
2000, 2029-2033
108 Chapter 4
16 Mvula, E , Schuchmann, M N, von Sonntag, C ,Reactions of phenol-OH-
adduct radicals Phenoxyl radical formation by water elimination vs
oxidation by dioxygen J. Chem. Soc, Perkm Trans. 2001, 2, 264-268
17 Mvula, E,von Sonntag, C, Ozonolysis of phenols in aqueous solution
Org. Biomol. Chem. 2003, /, 1749-1756
18 Mark, G, Hildenbrand, K, von Sonntag, C, in preparation
19 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Wat. Res. 1981, 15, 449-456
20 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water - II Dissociating organic compoundsWat. Res. 1983, 17, 185-194
21 Lee, C F In Principles and Applications of Water Chemistry, Faust, S D,
Hunter, J V, Eds, Wiley, New York, 1967, p 54-74
22 Rebenne, L M, Gonzalez, A C, Olson, T M
, Aqueous Chlormation
Kinetics and Mechanism of Substituted DihydroxybenzenesEnvironmental Science and Technology 1996, 30, 2235-2242
23 Well, I, Morris, J C, Kinetic studies on the chloramines I The rates of
formation of monochloramine, N-chloromethylamine and N-
chlordimethylamine J. Am. Chem. Soc. 1949, 71, 1664-1671
24 Abia, L, Armesto, X L, Canle, M , Garcia, M V, Santaballa, J A, Oxidation
of akphatic amines by aqueous chlorine Tetrahedron 1998, 54, 521-530
5 Enhanced Bromate Control DuringOzonation: The Chlorine-Ammonia Process
Buffle, M -O, Son]a Galk, Urs von Gunten
Environmental Science and Technology, 2004,38, 5187-5195
5.1 Abstract
Potentially caranogemc bromate forms during the ozonation of bromide-containing
waters. Some water treatmentfaahties have had to use ammonia addition andpH
depression to minimize bromate formation but these processes may prove to be
insufficient to comply with upcoming regulations. The chlorine-ammonia process
(Cl2-NH3), consisting of pre-chlonnation followed by ammonia addition prior to
ozonation, is shown to cause afourfold decrease in bromateformed when compared
to the ammonia-only process. Experiments revealed three key mechanisms:
(i) oxidation by HOCl of Br to HOBr and its subsequent masking by NH3 as
NH2Br; (a) decrease of HO' exposure through halogenation of Dissolved Natural
Organic Matter (DNOM) by HOCl and scavenging of HO' by NH2Cl;
(in) DNOM acting as a bromine sink after oxidation of Br to HOBr.
At an ozpne exposure of 6 mg/E-min andpH 8, conventional ozonation of Take
Zurich water spiked with 560 jUg/L Br formed 35 jUg/L Br03, whereas the
application of the Cl2-NH3 process resulted in 5 jUg/L Br03. AdditionalpH
depression to pH 6 further decreased bromate formation by a factor of 4.
Tnhalomethanes (THM) and cyanogen chloride (CNCl), that may form during
pre-chlonnation and monochloramination respectively, were well below regulatory
limits. The chlorine-ammonia process holds strong promise for water treatment
faahties struggling with a bromateformationproblem during ozonation.
5.2 Introduction
Ozonation is applied worldwide m the water industry as a disinfection
and oxidation treatment step (taste and odor removal, color removal,
iron and manganese oxidation, micropollutants degradation, etc.).
no Chapter 5
5.2.1 Ozone stability in water
The rate of aqueous ozone decay in natural water mostly depends on
the concentrations of dissolved natural organic matenal (DNOM),
carbonate ions, hydroxide ions and temperature. Ozone decay is
charactenzed by biphasic kmetics. The rate of the initial phase is very
high with half-kves m the order of seconds. The amount of ozone
consumed dunng this phase is referred to as ozone demand.
The second phase is well modeled with first order kmetics and exhibits
half-lives in the order of minutes to hours. This second phase is the
result of complex radical-type chain reactions initiated by hydroxide
ions and specific DNOM moieties (1,2). Hydroxyl radicals (HO') are
important products and chain earners of these reactions. The reaction
of HO' with DNOM leads to carbon centered radicals, which after
reaction with O2, can lead to the formation of superoxide (O2*).
Superoxide then reacts with ozone to generate more hydroxyl radicals
(chain reaction promotion). HO' can also be scavenged by compounds
(e.g. bicarbonate), which do not generate superoxide, thereby stabikzmg
ozone in the water (inhibition of chain reaction) (1,2).
5.2.2 Characterization of ozonation processes
Predictions of oxidation processes dunng water ozonation must
therefore take two main oxidants into account: ozone (O3) and
hydroxyl radicals (HO'). Elovitz and von Gunten (3,4) found that the
ratio (R«) of the hydroxyl radical exposure (HO' concentration
integrated over time: J[HO']dt) to the ozone exposure (J[Oj]dt) is
constant dunng the second phase of ozonation for a given set of water
quakty parameters. Under standard ozonation conditions, a R« of
approximately 10 8can be expected (4). An increase m pH, DNOM and
temperature, and a decrease m bicarbonate concentration result m an
mcrease in R^. Dunng the initial phase, however, R^ is not constant
The CI2-NH3 process for bromate minimization 111
and can be 10 times larger than dunng the second phase (3,4).
Taking Rct into consideration, the contribution of each oxidant (HO' or O3),
for the oxidation of a specific compound P can be expressed as:
MP]/P]o)=-{fco. -![HO']dt+/fo3 -j[03]dt} =-J[03]dt (kH0. -Rc+fcs) (1)
Clearly, if kos is much larger than kao- 'Rct (see nght hand side of eq 1),
the compound will be mostly oxidized by ozone, and vice versa.
5.2.3 Bromate formation
Ozone's abikty to oxidize bromide to bromate has been known and studied
as far back as 1942 (5). However, detailed mechanistic and kinetics
investigations were only initiated m the 1980's (6). Tnggered by a WHO
report that classified bromate as a potential carcinogen (7), the number of
pubkcations on bromate formation increased significantly m the 1990's.
Since then the complex pathway of bromate formation dunng ozonation
has been elucidated satisfactorily (8,9).
Scheme 1 gives an overview of the most important reactions. In a first step,
bromide is oxidized to Br' by HO' or to HOBr/OBr by O3. While for large
Rct (e.g. initial phase) the HO' pathway is important (> 40% of Br is oxidized
by HO' at Rct > 107), for typical Rct values (e.g. second phase) Br oxidation
by O3IS the prime reaction (96% of Br is oxidized by O3 at Rct = 10 ^ (9).
The product HOBr/OBr is therefore a key intermediate. HO' reacts with
both speaes (HOBr and OBr) with similar rates, while O3 only reacts with
OBr. At pH 7-8, HOBr being the dominant speaes pKa = 9 (8), all
constants mentioned in this article refer to 20°Q, the HO' pathway is
favored, resulting in the formation of oxidobromtne radical (BrO'). Br' also
forms BrO' through reaction with O3 or forms HOBr through a multiple
steps reaction with Br (at Br = 40 ug/L and O3 = 1 mg/L, 40% of Br' form
BrO' and 60% form HOBr). BrO' disproportionates into BrO and Br02
and the lattens readily oxidized by ozone to bromate (9).
112 Chapter 5
Scheme 1. Key reactions involved in bromate formation, (a)Bromate formation during conventional ozonation, adapted from (8).
(b) Reactions induced by pre-chlorination and ammonia addition
during the CI2-NH3 process.
The CI2-NH3 process for bromate minimization 113
5.2.4 Bromate minimization
As a result of the USEPA and EU setting bromate drmkmg water
standards at 10 ug/L, control strategies to minimize bromate
formation have become necessary for some treatment facilities.
Based on improved mechanistic and kinetic understanding, two
mam bromate control strategies have been applied to drmkmg
water ozonation: pH depression and ammonia addition (10,11).
The effects of pH depression and ammonia addition can be well
explained with the above-described mechanisms. pH depression
displaces the HOBr/OBr equilibrium further to HOBr, slowing
down the oxidation by ozone. In addition, it lowers the Rct value,
decreasing the rate of all HO' based oxidation processes.
Ammonia addition masks the key intermediate HOBr as NH2Br.
NH2Br then reacts slowly with ozone to form NO3 and Br
(k = 40 M h \ ti/2 > 15mm for 1.5 mg/L O3).
Applications of the pH depression and NH3 addition processes
result m bromate reduction of roughly 50% (11). With facilities
required to achieve several log inactivation of Cryptosporidium
parvum oocysts (C. parvum) and a continuous pressure from
regulators to further lower bromate standards, efficiencies of
those processes may well be found insufficient.
It is the goal of this work to elucidate the underlying principles of
a new bromate minimization strategy (12,13) that includes a pre¬
chlorination step prior to ammonia addition and ozonation
(the Cl2-NH3 process).
114 Chapter 5
5.3 Materials and Methods
5.3.1 Standards and Reagents
Milk-Q water with a resistivity above 18 MQ-cm was used to prepare all
aqueous solutions. All reagents were analytical grade. The aqueous
ozone stock solution (1.1-1.3 mM) was prepared by sparging a
5% ozone/oxygen gas mixture through ice-bath cooled Milli-Q water.
The NaOCl stock solution (Riedel-de-Haen, 6% active CI) was found
to contain significant amounts of bromate (mole fraction ~0.11%); this
is m agreement with findings of Weinberg et al. (14). Indigo solutions
were used: (l) to quench ozone for bromate analysis (0.2 mM Indigo
tnsulfonate) and (u) to quantify ozone concentration (0.113 mM Indigo
solution with 1% concentrated H3PO4 and 0.5 g/L malomc acid to
quench HOCl before its reaction with mdigo). The ABTS (2,2-azmo-
bis(3-ethylbenzothiazolme)-6-sulfonate) reagent for the analysis of
chlorine species (HOCl, NH2C1) contained 20% (v/v) lg/L ABTS
solution, 60% (v/v) 0.5 M phosphate buffer at pH 6 and 20% (v/v)
1 mM potassium iodide. Tnhalomethanes (THM: tnchloromethane,
tnbromomethane, bromodichloromethane and dibromochloromethane)
were obtained from Fluka Chemie GmbH.
5.3.2 Natural Waters
All expenments were performed m natural waters. Lake Zunch water
(mesotrophic) was sampled at the raw water intake of a dnnkmg water
treatment plant, 30 meters below the lake surface. The water quality
parameters are very constant and lead to reproducible conditions
throughout the year. High DOC and high initial NH3 expenments were
performed with Lake Greifensee water (eutrophic), sampled m an
effluent nver 200 meters downstream from the lake.
The CI2-NH3 process for bromate minimization 115
Table 1. Water quality data of Lake Zurich and Lake Greifensee water
DOC
[mgC/L]Alkalinity
[mM]
NH3
["g/L]
Br
["g/L]PH
H
Lake Zurich water
Lake Greifensee water
17
36
26
38
20
170
10
20
75
78
Lor expenments at pH 6 and pH 7, 5 mM phosphate buffer was used.
To avoid calcium phosphate precipitation, expenments at pH 8 were
buffered with 2-5 mM borate buffer. Higher borate buffer
concentrations were found to somewhat interfere with bromate analysis
by ion chromatography, so the lowest possible buffer concentrations
were used. pH was ad|usted for all buffers by adding H2SO4 or NaOH.
5.3.3 Analytical Methods
HOBr stock solution was quantified at pH 11 (as OBr) and 329 nm with
6 = 332 M ion1 (15). HOQ stock solution was quantified at pH 6
(as HOQ) and 230 nm with 6 = 100 M ^m1 (16). HOQ m concentrations
of 1 to 20 uM was analyzed with ABTS (17). Aqueous ozone stock
solution was quantified at 258 nm with 6 = 3000 M icm1. Ozone in
concentrations between 0.2-50 uM was analyzed with the Indigo
method (18). Tnhalomethanes were analyzed with headspace gas
chromatography on a Lisons GC8000, using an electron capture detector
(ECD) and a DB-5 column (19). Samples were transferred into GC glass
vials with Teflon caps immediately after sampling and measured the same
day. Samples were preheated 15 minutes at 60°C, 1 mL headspace gas was
sampled and in|ected (spMess) with the following temperature program:
31°C for 3 mm, l°C/min to 44°C, 15°C/min to 219°C for 2 mm.
The respective retention times for CHCI3, CHBrCL., CHB^Cl and CHEfe
were 4, 7, 12 and 18 minutes. Entire sequences lasted many hours, a dnft
due to concentration changes in the vials headspace as a function of time
reached up to 20% (first versus last control). Live point cakbrations were
116 Chapter 5
executed for all THM analyses: 0.1, 0.5, 1, 5 and 10 ug/L. Quantification
limit was 0.1 ug/L for CHCL, 0.03 ug/L for CHBrCl2, 0.1 ug/L for
CHBr2Q and 0.2 ug/L for CHBr3 based on 3 times the standard deviation
of the basekne. Bromate was analyzed with ion chromatography and UV
detection after a post-column reaction (20). A Kronlab LDP-5 high
preasion syringe pump was used for dekvery of the post-column reagent,
yielding better cakbration factor reproduabikty. Six point cakbrations were
performed for all bromate analyses: 0.2, 0.5, 1, 5, 10 and 20 ug/L
(100 ug/L for the high [Br] expenment). Quantification limit was
0.15 ug/L, based on 3 times the standard deviation of the basekne.
Bromate contained m the stock NaOCl solution were accounted for dunng
bromate analysis by subtracting the bromate concentration m the blank
fiom the measured value. To assess the magnitude of uncertainties due to
analytical and experimental errors, expenments represented m Fig 2 and 3
were repeated 3 times. The 95% confidence interval of the repkcate senes
with the largest scatter was calculated and normakzed to its mean
concentration. The obtained value (20%) was then used for the error bars
exempkfied in Figure 5.3. For the sake of clanty error bars were left out of
the other figures. Bromide was measured with IC and conductivity detection
(20), and five point cakbrations were earned out: 5,10, 50,100 and 200 ug/L
(1000 ug/L for the high [Br] experiment). Quantification limit was 4 ug/L,
based on 3 times the standard deviation of the basekne. HOBr and NH2Br
are reduced on the column so that bromide measured by IC is the sum of
Br, HOBr and NH2Br. However, the bromine that reacts with DNOM is
not reduced on the column. Organically bound bromine concentration
could therefore be calculated by subtracting the measured [Br] and [Br03 ]
fiom [Br]0. Hydroxyl radicals (HO') were quantified indirectly by analyzing
para-chlorobenzoic acid (pCBA) using HPLC separation and UV detection
at 240 nm according to (3). The quantification limit was 10 ug/L (0.06 uM).
1 uM^CBA was added to the water to be ozonated and its decrease yielded
the hydroxyl radical exposure (with /èHo-^CBA = 5-109 M h \ (3)).
The CI2-NH3 process for bromate minimization 117
5.3.4 Experimental Setup
AH kinetics experiments were performed at 20°C with half kter amber glass
bottles capped with a dispenser. They were filled with 500 mL buffered
natural water and spiked with bromide and/or ammonia to the desired
concentrations. Hypochlorous aad and/or ammonia were added as
concentrated stock solutions (< 1 mL of ~10 mM). 10-20 milliliters of the
aqueous ozone stock solution were added with a gas-tight Hamilton syringe
through a capillary tube drilled m the dispenser's cap under sttrnng condition.
As soon as inaction was ended (< 4 s) the bottle was inverted three times.
The dispenser was purged twice and 6 mL samples were taken at 0.5,1, 2, 5,
10,20,40 and 60 minutes. The experimental sequence is shown in Figure 5.1.
CJ,NH3 Process
AW3 Process
Figure 5.1. Expérimental sequence natural waters were spiked,buffered and adjusted for pH Following chlorine addition with 5
minutes contact time, ammonia and ozone were added All experiments
were performed at 20°C
A typical bromate formation kinetics expenment required up to 4 senes
of test tubes to be prepared: (1) bromate analysis (2 mL of 0.2 mM
Indigo reagent), (u) ozone analysis (2 mL of 0.113 mM Indigo reagent),
(in) THM analysis (samples taken at 0 and 60 minutes; 0.2mM Indigo
or thiosulfate reagent) and (w) pCBA analysis (300 uM sulfite reagent).
An independent senes of THM formation expenments was performed.
These were done m smaller reaction bottles (40 mL) using a Hamilton
synnge as a sampkng device instead of a dispenser. Samples were taken
at 0 and 60 minutes and analyzed for THM and bromate.
118 Chapter 5
5.4 Results and Discussion
5.4.1 Effect of pre-chlorination
Dunng the Cl2-NH3 process pre-chlonnation is followed by ammonia
addition. The following three reactions have to be considered
([NH3]>[HOCl]>[Br ]):
HOQ + Br ->HOBr + Cl k = 1550 M is1 (21) (2)
pKaHoci = 7.5 (21)
HOBr + NH3 -> NH2Br + H20 k = 8407 M h 1 (22) (3)
pKanoBr = 9 (8)
NH2Br + 303^N03+Br +3O2 + 2H+/è = 40M1s1 (22) (4)
In reaction 2, naturally occurring bromide is oxidized by
hypochlorous acid to hypobromous acid. This is followed by
reaction 3 where NH3 reacts with HOBr to form monobromamme.
The rate constant for monobromamme formation being very high,
the reaction time necessary to complete this step is insignificant
(ti/2 < ms for NH3 = 10 uM). NH3 also reacts with free chlorine to
form monochloramme. However, this reaction does not affect
monobromamme formation as ammonia is added m excess of
HOCl and the rate of monobromamme formation is 20 times
higher than that of monochloramme formation (eq 3 and 6).
Subsequent addition of ozone, m reaction 4, leads to the sluggish
oxidation of monobromamme to nitrate and bromide (ti/2 > 15 mm
for O3 = 1.5 mg/L). Monobromamme also transforms slowly to
dibromamme, which is oxidized by ozone at a much lower rate than
NH2Br NHBr2 does not decay further over practical ozonation
times (10 - 20 mm) (23).
The CI2-NH3 process for bromate minimization 119
Hence, under practical conditions most of the bromine remains
masked as bromamme for the duration of the ozonation process.
Some recycled Br (reaction 4) may be oxidized again by ozone to
HOBr. However, due to excess of ammonia it is quickly masked as
NIrfcBr. Alternatively, some of the recycled bromide can be oxidized
by HO' to Br' and eventually to bromate (Scheme la). However in
the secondary phase of ozonation, this pathway plays a minor role
in bromate formation (11).
Based on reactions 2-4, the key improvement of the Cl2-NH3
process over the conventional bromate minimization processes
(pH depression, NH3 addition) is the hindering of Br oxidation to
Br' by HO' during the initial phase of ozonation (Scheme la).
Reactions 2-4 however do not exhaustively descnbe the Cl2-NH3
process. Additional key mechanisms positively impacting the
efficacy of the process are discussed in the following sections.
Figure 5.2a shows the calculated bromide conversion to HOBr for
HOCl doses of 5, 10, 15 uM HOCl (0.35, 0.7, 1.05 mg/L Cl2) in
Lake Zurich water. Bromide conversion was calculated based on the
measured HOCl exposures (mset, Fig 2a) according to:
ln([Br]/[Br]0) = - fcoaBrJrHOOJdt (5)
At initial HOCl concentrations of 5, 10 and 15 uM, and a practical
reaction time of 5 minutes, ~30%, ~60% and ~75% of the
bromide is oxidized to hypobromous acid, respectively.
Although the purpose of the pre-chlonnation step would be to
oxidize bromide completely before it is masked by ammonia,
pre-chlonnation can also lead to the formation of trihalomethane.
Thus it is preferable to minimize free chlonne exposure as much as
possible (see section on THM formation).
120 Chapter 5
Ozone exposure [mg/L mm]
Figure 5.2. C12-NH3 process Effect of pre-chlorination on HOBr and
Br03 formation in Lake Zurich water at 20°C and pH 8 (a) Free
bromine formation calculated with measured free chlorine exposure
Inset Chlorine decrease measured for 5, 10, 15 uM HOCl (0 35, 0 7,
1 05 mg/L CL) (b) Measured Br03 formation plotted as a function of
ozone exposure [Br ]q = 90 ug/L (1 1 uM), pre-oxidaüon for 5 minutes
with 0, 5, 10, 15 uM HOCl (0 35, 0 7, 1 05 mg/L Cl2), followed byaddition of 300 ug/L (18 uM) NH3 and 1 5 mg/L (31 uM) ozone
The CI2-NH3 process for bromate minimization 121
Figure 5.2b shows the kinetics of bromate formation in the Cl2-NH3
process for vanous HOCl doses and constant NH3 dose
(with [NH3]>[HOCl]). The benefit of the Cl2-NH3 process over NH3
addition only (astensks) is significant. With an addition of 10 uM
HOCl (tnangles) and at an ozone exposure of 6 mg/L-mm
(i.e. 2-log C. parvum inactivation at 20°C (24)), bromate formation
decreased by a factor of 4 compared to NH3 addition only (astensks).
NH3 addition only decreased bromate formation by approximately a
factor of 2 compared to conventional ozonation (data not shown, for
more details see (1 /)). Figure 5.2b also shows that for this specific
natural water the gam m increasing HOCl concentration beyond 10 uM
is small because bromide conversion is already quite high (~60%).
Nevertheless, small amounts of bromate are still formed through
HO'-mduced oxidation of the non-converted bromide, and of the
bromide recycled dunng ozonation of NH2Br (see earker discussion).
5.4.2 Effect of ammonia addition
Based on the above results, a 10 uM HOCl dose was fixed and the
effect of varying ammonia addition was investigated. In Figure 5.3,
bromate formed at an ozone exposure of 6 mg/L-mm is plotted
against the ammonia dose. A 10-fold decrease in bromate formation
can be observed between pre-chlonnation without NH3 addition and
pre-chlonnation with 300 ug/L NH3 addition (solid tnangles). The
single sokd square symbol also shows bromate formed dunng
conventional ozonation (no CI2, no NH3) for the same water and ozone
exposure. In the process with ammonia addition only, Pinkernell and
von Gunten (11) had observed a 2-fold decrease m bromate formation.
They also noted that only little gam was obtained by increasing the
NH3 concentration beyond 188 ug/L (11.1 uM) (11). In Figure 5.3, a
levekng off can be observed at an ammonia dose of 300 ug/L
(17.6 uM). This is comparable to results in (11) because after 5 minutes
122 Chapter 5
of chlormation, 8 uM free chlonne and bromine are still in solution,
the addition of 17.6 uM NH3 then generates 8 uM {[NH2C1] +
[NH2Br]} and leaves 9.6 uM (164 ug/L) unreacted NH3 which is very
close to the 188 ug/L findings m (11). It is also noteworthy that when
chlonne is dosed m excess of ammonia, it can promote bromate
formation by pre-formmg the key intermediate HOBr (compare solid
square and tnangle at NH3 = 0 m Fig 3). However, this effect might be
partly compensated by changes m the overall ozone reactivity towards
NOM due to the pre-chlonnation step (see below).
14 n
12 -
10l
_1
8 -
0)
E0
m
6 -
4 -
2 -
[HOBr]+[HOCI]=8 |JM ^4-_>
~~~ "* -^
° '
Ï36' ' ' ~,
0 100 200 300 400 ^/L
0 598
118 176 235 |jM
Ammonia
Figure 5.3. CI2-NH3 process Effect of ammonia dose on bromate
formation in Lake Zunch water at 20°C and pH 8 Bromate formed at
ozone exposure of 6 mg/L-min plotted as a function of NH3 doses of 0,
100, 200, 300 and 400 ug/L (0, 6,12,18, 24 uM) [Br]0 = 90 ug/L (11 uM)and [NH3]o = 20 ug/L (1 2 uM) Pre-oxidaüon for 5 minutes with 10 uM
HOQ (0 7 mg/L CI2), followed by addition of ammonia (see doses above)and 1 5 mg/L (31 uM) ozone The single sohd square symbol represents
bromate formed during conventional ozonation (no Q2, no NH3) for the
same water and ozone exposure as above The error bars represent a 20%
uncertainty calculated by normalizing the 95% confidence interval of the
senes with the largest scatter to its mean concentration (n—3)
The CI2-NH3 process for bromate minimization 123
5.4.3 Effect of naturally occurring ammonia
Based on the above-presented mechanism (eqs 2-4), naturally occurnng
ammonia (i.e. NH3 present m the water before the chlonnation step)
should significantly reduce the efficiency of the Cl2-NH3 process.
Comparing equations 2 and 6, NH3 reacts with HOCl to form NH2CI
before HOCl oxidizes Br to HOBr. Monochloramme then reacts so
slowly with bromide to monobromamme that reaction 7 is insignificant.
HOCl + NH3 -» NH2Q + H20 k = 4.2-106 M h 1 (25) (6)
NH2C1 + Br ->NH2Br + Q >è = 0.014M1s1 (26) (7)
To investigate the limitation of the Cl2-NH3 process in ammonia-
contaming waters, expenments were conducted with 170 ug/L
pre-spiked ammonia (10 uM). According to eq 6 and 7, no decrease in
bromate formation would be expected when HOCl is dosed below
the pre-spiked NH3 concentration.
Figure 5.4a shows the effect of HOCl on bromate formation for
doses between 0 and 15 uM. Even for HOCl concentrations 4 times
below the pre-spiked NH3 concentrations, bromate formation is
significantly reduced. This indicates that the mechanism proposed
with reactions 2 - 4 is incomplete.
Figure 5.4b shows the effect of varying chlonne doses on the
oxidation of a HO' probe (para-düotobenzoic acid) dunng the
subsequent ozonation step for the same senes of expenments. An
increase in the HOCl dose, leads to slower ^CBA elimination. This
indicates a lowered Rct, i.e. a decrease in HO' exposures at a given
ozone exposure. This explains the bromate formation, which for a
given ozone exposure, increases at decreasing HOCl concentrations
due to the increased overall oxidant exposure (O3 + HO').
124 Chapter 5
a; 2 -
1
15 uM HOCl
Ozone exposure [mg/L mm]
Ozone exposure5 [mg/L mm]
0 0 00125 0 0025 0 00375 0 005 0 00625 IM sl
-!- -0 4-
?m
oQ- -0 6-
<m
^ -0 8-
Figure 5.4. C12-NH3 process Effect of pre-chlorination dose in
NH3- and Bi -spiked lake Zunch water at 20°C, pH 8,
[Bi]o = 90 ug/L (11 uM) and [NH3]o = 170 ug/L (10 uM)
(a) bromate formation and (b) pCBA decrease as a function of ozone
exposure The water was pre-oxidized with 0, 2 5, 5, 15 uM HOCl
(0 175, 0 35, 1 05 mg/L CL) during 5 minutes, followed by the
addition of 300 ug/L (18 uM) NH3 and 1 5 mg/L (31 uM) ozone
The CI2-NH3 process for bromate minimization 125
Cb [mg/L]
0 175 0 35 0 525 0 7 0 875 1t
25 5 75 10
HOCl [uM]
<
1 25- (b)
0 75- /•
05-/•
0 25-
0-—•—^
HO exposure / HO exposure a
Figure 5.5. CI2-NH3 process Importance of HO" reactions on
bromate formation, with same experimental conditions as in
Figure 5 4 (a) Initial HO" exposure (at ozone exposure of
1 mg/L-min) as a function of the pre-chlorination dose normalized bythe initial HO" exposure when no pre-chlorination is performed
(b) Natural log of bromate formed at various pre-chlorination doses
normalized by bromate formed at highest HOCl dose (lowest bromate
formation), as a function of the initial HO" exposure ratios
126 Chapter 5
Figure 5.5a shows the initial HO' exposures (at ozone exposure of
1 mg/L-mm, t~lmm) normalized to the case with no
pre-chlonnation plotted as a function of the chlorine dose.
At 10 liM HOCl (1.05 mg/L Cl2), the HO' exposure is roughly half
of the HO' exposure when no pre-chlonnation is performed. Again,
a diminishing return can be observed with increased HOCl dose.
Figure 5.5b demonstrates the relationship between the decrease m
bromate formed (at an ozone exposure of 1 mg/L mm) and the
decrease m initial HO' exposure due to the addition of HOCl. The
strong dependency of bromate formation on HO' exposure is
consistent with HO' being a key pathway for bromide oxidation
dunng the initial phase of ozonation as well as when large NH3
concentrations mask HOBr.
Two mechanisms could explain a decrease of HO' exposure after
addition of HOCl m NH3-contammg waters: (1) HOCl or/and
NH2CI oxidizes specific moieties of the DNOM and reduces their
reactivities towards ozone which subsequently hinders the
generation of HO', (11) HOCl or its oxidation or substitution
products act as HO' scavengers.
For mechanism (1) to be plausible m ammoma-contammg water
either HOCl must react with specific DNOM moieties at higher
rates than with ammonia, or NH2CI oxidation of certain DNOM
moieties must be significant during the 5 minutes pre-oxidation step.
HOCl is known to react rapidly with phenols (e.g. resorcmol
k — 3.4-104 M H 1at pH 8, (27)). Phenolic groups concentrations m
lakes DNOM are estimated to be on the order of 6 uM (28). Despite
its relatively high rate, HOCl reaction with phenols (e.g. resorcmol) is
still 100 times slower than with NH3 (k = 4.2406 M h 1).
The CI2-NH3 process for bromate minimization 127
Thus, taking their respective concentrations into account, the
likelihood of significant phenolic oxidation by HOCl pnor to
NH2CI formation is small.
Few data are available on NH2C1 oxidation of phenolic compounds.
Resorcmol reaction rate with NH2CI was determined in this study as
-1.5 M h 1 (pH 7.8, 20°C, 2 mM borate buffer, 100 uM NH2C1 and
10 uM resorcmol). Kirankumar and Haas found a slightly higher
rate constant for phloroacetophenone (k = 4.5-6.9 M as 1, (29)),
which can be expected because phloroacetophenone is further
activated by an additional hydroxyl group. Even assuming an
excessive reaction rate constant of 10 M 1s 1 with phenolic groups,
NH2CI could not account for more than 3% of the phenolic
moieties' oxidation dunng the 5 minute pre-oxidation step.
HOCl is also known to react rapidly with amines (e.g. glycine
S-lOTMis1 (30)). Total hydrolysable ammo acid (THAA)
concentrations m rivers and lakes are estimated to be m the range
of 2 - 26 uM, with glycine representing -20% of THAA (28).
In contrast to reactions with phenols, the reactions between HOCl
and organic amines (e.g. glycine) occurs 10 times faster than the
reaction between NH3 and HOCl when glycine and ammonia are at
equimolar concentrations.
Based on the above-estimates, a significant fraction of HOCl can
therefore be expected to react with nitrogenous compounds of the
DNOM. The halogenation of organic amines through
pre-chlonnation is likely to decrease their oxidation rate by ozone,
which typically reacts very rapidly with deprotonated amines (1,2).
This may result m a slower initial ozone decay and smaller initial
HO' generation.
128 Chapter 5
Mechanism (u), the scavenging effect of HOCl or its oxidation
products on HO', can be evaluated by estimating the fraction (fko-) of
HO' reacting with the vanous chemical species m the water:
/ {A) =
[A] ' kA-HO (8)> HO
V >TT^n
Z.lJX.^kX,-HO-
where [A] is the concentration of a compound A for which the
fraction is to be calculated; k^ ho is the second order rate constant
for the reaction of A with HO' and Xt is the ith of n compounds
involved m the scavenging of HO'.
The speciation of the key compounds (Br, NH2Br, HOBr, OBr,
HOCl, OC1, NH2C1, NH3, CO32, HCO3, NOM; for kA Ho values
see (2,9) after chlonne and ammonia addition (pnor to ozonation)
were computed using a kinetics solver (ACUCHEM (31)) and entered
m equation 8. With 10 uM (170 ug/L) pre-spiked NH3, 10 uM
(0.7 mg/L) HOCl and 18uM (300 ug/L) NH3 addition, a net decrease
of —7.5% m HO' available for Br oxidation was calculated (based on
/Wciho. = 5408 M^1, (32)) [Johnson et al. (33) found
/èNH2ciHo- = 2.8 409 Mis !. Using this rate constant, 27% of HO'
would be scavenged by NH2CI. In the present paper, however, we
used the rate constant found by Poskrebyshev et al. (32). For a more
detailed discussion see (32)\ Reaction of NH2C1 with HO' leads to
the formation of 'NHC1, which eventually produces NO (32).
In conclusion, the decrease m bromate formed dunng the Cl2-NH3
process m ammoma-contatnmg water is largely due to a lowered Rct, i.e.
a lowered HO' exposures for given ozone exposures. We propose that
this is due to a combination of pre-chlonnation of HO' generating
moieties m the DNOM and direct HO' scavenging by NH2CI. A study
is underway to investigate these hypotheses.
The CI2-NH3 process for bromate minimization 129
5.4.4 Effect of DNOM as a sink for bromine
In Figure 5.6, the fraction of bromine bound to organic matter (TOBr) is
plotted against pre-chlonnation dose. The fraction complement represents
the sum of bromide, the bromine speaes reduced to Br on the IC column
(NHzBr, NHBr2, HOBr/OBr) and bromate. TOBr increases with HOCl
dose and reaches 20% of the initial bromide concentration at 10 uM HOQ
when the initial NH3 concentration is low (open tnangles m Fig 6). THM
formation for a HOQ dose of 10 uM was also measured. Whereas 20% of
the bromine binds to DNOM, THM accounted for less than 5% of the
organically bound bromine (Br03 < 1%). For expenments where ammonia
is pre-spiked (rapid chloramine formation) the amount of bromine bound
to the organic matrix is significantly reduced (solid circles in Fig 6).
Cl2 [mg/L]
0 35 0 7
10 15
HOCl [MM]
Figure 5.6. C12-NH3 process Organically bound bromine plotted as a
function of the pre-chlorination dose in waters with différent initial
ammonia concentrations Both data series with Lake Zurich water at
20°C and buffered at pH 8, [Br ]o = 90 ug/L (1 uM), pre-chlorinatedwith various HOCl concentrations during 5 minutes, followed by the
addition of 300 ug/L (18 uM) NH3 and 1 5 mg/L (31 uM) ozone
Open triangles water with natural NH3 concentration of 20ug/L(1 2uM) Solid circles water pre-spiked with 170 ug/L (10 uM)
130 Chapter 5
Thus, an additional bromate minimization mechanism results from
DNOM acting as a bromine sink:
HOBr + DNOM = TOBr (9)
Even though reaction 9 leads to a reduction m bromate formation, total
brommated orgamcs, although not regulated, may still be a health concern.
5.4.5 Formation of trihalomethanes and cyanogen chloride
As discussed earlier, a possible drawback of the Cl2-NH3 process is
the formation of halogenated organic compounds either through
chlormation or through brommation of DNOM. Due to the
simultaneous presence of HOBr and HOCl, four tnhalomethanes
can be expected: CHBr3, CHB^Cl, CHBrCfe, CHCL- However, given
that the chlorine exposures required for the Cl2-NH3 process are
small and free bromine and chlorine are masked by NH3 as
bromamme and chloramme, low THM formation can be expected.
For Lake Zunch water spiked with 90 ug/L bromide and pre-
chlormated with 10 uM HOCl (0.7 mg/L Cl2) the sum of all THM
(TTHM) reached 3.5 ug/L after 1 hour. This is well below the
drmkmg water standards of 100 ug/L set by the EU (34) and
80 ug/L set by the USEPA (35).
In Figure 5.7, TTHM and bromate formation are plotted as a function
of the pre-chlonnation dose for waters containing very high bromide
concentrations ([Br]=590 ug/L). Dark columns show results fiom
expenments with Lake Zunch water, light columns with Lake
Greifensee water. Although Lake Greifensee water has more than twice
the DOC concentration of Lake Zunch water (3.6 vs. 1.7 mg/L), less
THM are formed. This is largely due to the high concentration of
naturally occurnng ammonia m Lake Greifensee water (10 uM NH3). If
less than 10 uM HOCl is added to Lake Greifensee water, free chlonne
The CI2-NH3 process for bromate minimization 131
is rapidly transformed into monochloramme. As a result, only minor
quantities of THM are formed because of the very slow reaction
between NH2C1 and DNOM.
An additional explanation for the low THM concentration in lake
Greifensee water is that when calculating the gravimetric sum of
THM (as required by the regulators), brommated compounds, due to
their high molecular weight, have a more important share than their
chlorinated counterparts (particularly in waters containing very high
bromide concentrations). The larger amount of NOM in Lake
Greifensee water leads to a competition between the HOCl / NOM
and the HOCl / Br oxidation reactions. It follows that less HOBr
and consequently less brommated THM are formed in Lake
Greifensee water than in Lake Zunch water. This explains why even
at equivalent initial free chlorine concentrations (taking naturally
occurring NH3 into account: 10 uM HOCl dose in Lake Zunch
water S 20 uM HOCl dose in Lake Greifensee water) the higher
DOC water shows less total THM formation.
DNOM composition is considered relatively similar for both lake
waters (algae-derived organic matter), so that variation in THM
formation should mostly depend on DOC concentration. While
more bromate is formed in Lake Greifensee water than in Lake
Zurich water (25 ug/L versus 20 ug/L at [HOCl] = 0), the relative
decrease in bromate formed as a function of chlorine dose is
similar for both waters. Figure 5.7 exemplifies the trade-off
between bromate minimization and THM formation. In the cases
presented here, even though high chlorine doses were applied
along with very high bromide concentrations and high DOC
concentrations, TTHM concentrations were far below the dnnkmg
water standards and an optimization was not necessary.
132 Chapter 5
Lake Greifensee water 3 6 mg/L DOC - 170 Mg/L NH3
I Lake Zunch water 1 7 mg/L DOC - 20 ug/L NH3
Figure 5.7. C12-NH3 process Bromate and sum of THM (TTHM)formed as a function of the pre-chlonnation dose for Lake Zurich water
and Lake Greifensee water Both waters buffered at pH 8, spiked with
590 ug/L Br pre-chlonnated for 5 minutes, followed by the addition of
300 ug/L NH3 and analyzed after 60 minutes of ozonation (complete
decay) with 1 5 mg/L O3 (L Zurich water) and 3 mg/L O3
(L Greifensee water) Exact ozone exposures were not determined for
each data point but pre-expenments showed that the above-doses
correspond to ozone exposures of 10 ± 1 5 mg/L-mm in both waters
It should also be noted that results from these expenments can be
considered conservative in terms of DBP formation. At pH 8, TTHM
are significantly higher than at pH 6 and show higher molar
concentrations than haloacetic acids (HAA9) (36).
Ozonation of DNOM-contammg water is known to form small
amounts of formaldehyde (CH2O) (37). Pedersen et al. (38)
mvestigated the formation of cyanogen chloride (CNCl) dunng
chlorammation of formaldehyde-contammg water. Applying the
mechanisms and rate constants published m (38), the potential
The CI2-NH3 process for bromate minimization 133
formation of CNCl was modeled using ACUCHEM (31). With CH20
and NH2C1 concentrations of 30 ug/L (1 uM), and 540 ug/L (10 uM)
respectively, the concentration of CNCl was 7 ug/L after 15 minutes,
well below the 70 ug/L standard proposed by WHO (39). The CNCl
concentration estimated above represents a worst case scenario
because key intermediates m the formation of cyanogen chlonde are
ammo compounds (e.g. N-chloroammomethanol), which are known to
be readily oxidized by ozone when deprotonated. The actual
concentration of cyanogen chlonde is therefore expected to be well
below the value of 7 ug/L presented here.
5.4.6 Effect of pH
As mentioned earker, pH depression is an effective technique to decrease
bromate formation m conventional ozonation processes (11). With the
Cl2NH3 process the effect of pH depression is amplified. As pH is
depressed, the HOC1/OQ equilibnum (pKa = 7.5) is shifted toward HOCl,
which is the stronger oxidant. As shown in Scheme 1, the oxidation of Br
to HOBr occurs 106 times faster with HOQ than with OQ. It can
therefore be expected that reactions 2 and 9 will be strongly enhanced by
lowenng the pH. Most experiments presented m this study were performed
at pH 8. The results can therefore be interpreted as conservative.
In Figure 5.8, three bromate formation expenments are shown for pH 8, 7
and 6. When pH is depressed in parallel to applying the Cl2NH3 process,
the amount of bromate formed at pH 6 and at an ozone exposure of
6 mg/L-mtn is roughly 5 times smaller than at pH 8. At pH 6, bromate
concentration (~ 0.4 ug/L) remains near the quantification limit of the
analytical method up to an ozone exposure of 25 mg/L-mm. As a
remmder, conventional ozonation of the same water at pH 8 leads to
10 ug/L bromate at an ozone exposure of 6 mg/l-mtn (see solid square in
Fig 3), this is a factor 40 above the data presented m Fig 8 (open tnangles).
134 Chapter 5
When pH is depressed during conventional ozonation, the amount
of bromate formed at pH 6 is only about a factor of 2 smaller than
at pH 8 (11). Hence, pH depression has a synergistic effect on the
efficiency of the Cl2-NH3 process.
Although lowering the pH is not an economical solution for water
treatment facilities with high alkalinity waters, applying it
simultaneously to the Cl2-NH3 process may be recommended for
very difficult cases (e.g. very high bromide concentrations and
ozone exposures).
06-
04-
02
pH8
Ozone exposure [mg/L mm]
Figure 5.8. C12-NH3 process Bromate formation kinetics in Lake
Zurich water buffered at pH 6, 7 and 8 [Br]0 = 90 ug/L (1 uM),
pre-chlorinated with 10 uM HOCl, addition of 300 ug/L NH3 and
1 5 mg/L ozone
The CI2-NH3 process for bromate minimization 135
5.4.7 Implications for water treatment facilities
Based on our findmgs, the Cl2-NH3 process for bromate mmimization
can be charactenzed by 3 main features:
1. Oxidation of bromide by chlonne and subsequent formation of NH2Br
ü. Decrease of hydroxyl radical exposure due to a reduction in HO' generation
through DNOMpre-halogenation and HO' scavenging byNH2Q
ill. Incorporation of bromme mto the organic matrix
In Scheme 1, the reactions occurring during the Cl2-NH3 process
have been summarized graphically and combined with the
bromate formation mechanisms. The relative importance of each
mechanism varies depending on the water quality parameters
([Br], [NH3], [DOC] and pH) and on the process parameters
(CI2 exposure and NH3 dose). For water with low natural
concentrations of ammonia and DOC, mechanism (1) is
important. For water containing high natural concentrations of
ammonia and DOC, mechanism (11) is key. For water with high
DOC and low ammonia concentrations all three mechanisms are
important. To investigate the efficacy of the Cl2-NH3 process on
water containing a very high bromide concentration, experiments
were performed at pH 8 with Lake Zurich water spiked with
560 ug/L (6 uM) Br.This corresponds to the 99.6 percentile of
the 500 US water treatment plants entered m the ICR Database (40).
In Figure 5.9 bromate formation during conventional ozonation,
after ammonia addition and after the chlorine ammonia process
are shown. For conventional ozonation (solid squares) at an ozone
exposure of 6 mg/L-mm, 35 ug/L bromate are formed, 3.5 times
above the regulatory limit. Solid triangles represent ozonation
with the ammonia process (400 ug/L NH3). A clear improvement
136 Chapter 5
is seen with regard to conventional ozonation but the regulatory
limit is already exceeded at the first data point (ozone exposure of
1 mg/L-mm). At an ozone exposure of 6 mg/L-mm, the ammonia
process results in 18 ug/L bromate. The 50% reduction observed
previously can therefore be confirmed in this system (77).
The open circles represent bromate formed with the Cl2-NH3 process.
A pre-chlonnation dose of 16 uM (1.12 mgCfe/L) was applied. To
insure that HOX was masked as NH2X, and that an excess in
ammonia remained in the system, 400 ug/L (24 uM) NH3 were added.
The maximum bromate concentration is approximately 5 ug/L, a
factor of 2 under the regulatory limit.
40 n
[Bromide]o = 560 ug/LConventional ozonation (15 mg/L O 3)
30-
2 20-
1o
m
Drinking Water Standard
C /2 NH3 process (112mg/L CI2 400pg/INH1)
0 2 4 6 8
Ozone exposure [mg/L mm]
Figure 5.9. C12-NH3 process Bromate formation kinetics in Lake
Zurich water spiked with 560 ug/L Br,buffered at pH 8 and 20°C The
ozone dose was 1 5 mg/L for all expenments Solid squares indicate
conventional ozonation Solid triangles indicate ozonation with the NH3
process (400 ug/L NH3) Open circles indicate ozonation with the
C12-NH3 process (16 uM HOCl (1 12 mgCl2/L) and 400 ug/L NH3)Dashed line indicates the Br03 drinking water standard of 10 ug/L
NH3 process (4O0pg/L NH3)
The CI2-NH3 process for bromate minimization 137
It should be noted that the HOCl and NH3 doses were chosen based
on knowledge from expenments with lower bromide concentrations. If
required, further decrease in bromate formation could be obtained by
optimizing the doses for this specific water. Moreover, additional pH
depression would reduce bromate concentration to almost undetectable
levels. This study shows that the Cl2-NH3 process has the potential to
reduce bromate formation to levels well below the dnnkmg water
standards of 10 ug/L in waters containing very high concentrations of
bromide. It represents a promismg strategy for water treatment
facilities dealing with such problematic waters.
5.5 Acknowledgments
We thank Suez Environnement for financial support; Isabelle Baudm,
Auguste Bruchet, Zdravka Do-Quang, Mane-Laure Janex and Jean-
Michel Lamé for fruitful discussions; Jakov Bolotm, Pascal Jaeggi, Gun-
Young Park and Lisa Salhi for laboratory assistance; Marc Huber and
Gretchen Onstad for insightful comments.
5.6 References
1 Hoigné, J In The Handbook of Environmental Chemistry, Hrubec, J, Ed,
Springer Verlag, 1998, Vol 5, pp 83-141
2 von Gunten, U, Ozonation of drinking water Part I Oxidation kinetics
and product formation Wat. Res. 2003, 37, 1443-1467
3 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios duringozonation processes I The Rct concept O^one Sa. Eng. 1999, 21, 239-260
4 Elovitz, M S, von Gunten, U, Kaiser, H-P, Hydroxyl radical/ozone
ratlos during ozonation processes II The effect of temperature, pH,
alkalinity and DOM properties O^one Sa. Eng. 2000, 22, 123-150
5 Taube, H ,Reactions in solutions containing ozone, hydrogen peroxide,
H+ and Br- The specific rate of reaction 03+Br -> J. Am. Chem. Soc.
1942, 64, 2468-2474
6 Haag, W R, Hoigné, J, Ozonation of bromide-containing waters
kinetics of formation of hypobromous acid and bromate Environ. Sa.
Technol. 1983, 77,261-267
138 Chapter 5
7 WHO, IARCMonographs: Evaluation of Carcinogeinc Risks to Humans 1990, 52
8 von Gunten, U, Hoigné, J, Bromate formation during ozonation of
bromide-containing waters interraction of ozone and hydroxyl radical
reactions Environ. Set. Technol. 1994, 28, 1234-1242
9 von Gunten, U, Ozonation of drinking water Part II Disinfection and
by-product formation in presence of bromide, iodide and chlorine Wat.
Res. 2003, 37, 1469-1487
10 Glaze, W H, Weinberg, H S
, Cavanagh, J E, Evaluating the formation
of brominated DBPs during ozonation Jour. AWWA 1993, 85, 96-103
11 Pinkernell, U, von Gunten, U, Bromate minimization during ozonation
Mechanistic considerations Environ. Sei. Technol. 2001, 35, 2525-2531
12 Hulsey, R A, Neemann, J J , Zegers, R E
, Rexing, D J, Water treatment
using ozone and having a reduced likelihood of bromate formation from
bromides found in the water US Patent application 2003
13 Neemann,J J, Hulsey, R A, Rexing, D J, Wert, E, Controlling bromate formation
dunng ozonation with chlonne and ammonia. Jour. AWWA 2004, 96, 26-29
14 Weinberg, H S, Delcomyn, C A, Unnam, V, Bromate in chlorinated
drinking waters Occurrence and implications for future regulationEnviron. Sei. Technol. 2003, 37, 3104-3110
15 Troy R C, Margerum, D W, Non-metal redox kinetics hypobromiteand hypobromous acid reactions with iodide and with sulfite and the
hydrolysis of bromosulfate Inorganic Chemistry 1991, 30, 3538-3543
16 Soulard, M, Bloc, F, Hatterer, A, Diagrams of existence of chloramines and
bromamines in aqueous solution J.C.S. Dalton Transactions 1981, 12, 2300-2310
17 Pinkernell, U, Nowack, B, Gallard, H ,von Gunten, U, Methods for the
photometric determination of reactive bromine and chlorine species with
ABTS Wat. Res. 2000, 34, 4343-4350
18 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Wat. Res. 1981, 15, 449-456
19 Gallard, H, von Gunten, U, Chlonnauon of natural organic matter
kinetics of chlonnauon and of THM formation Wat. Res. 2002, 36, 65-74
20 Salhi, E, von Gunten, U, Simultaneous determination of bromide,
bromate and nitrite in low ug 1-1 levels by ion chromatography without
sample pretreatment Wat. Res. 1999, 33, 3239-3244
21 Kumar, K, Margerum, D W, Kinetics and mechanism of general-acid-assisted oxidation of bromide by hypochlorite and hypochlorous acid
Inorg. Chem. 1987, 26, 2706-2711
22 Haag, W R, Hoigné, J, Bader, H, Improved ammonia oxidation byozone in the presence of bromide ion during water treatment Wat. Res.
1984, 18, 1125-1128
The CI2-NH3 process for bromate minimization 139
23 Lei, H , Mannas, B J, Minear, R A,Bromamine decomposition kinetics
in aqueous solutions Envir. Sa. Technol. 2004, 38, 2111-2119
24 Finch, G R, Belosevic, M, Controlling Giardia spp and
Cryptosporidium spp in drinking water by microbial reduction processes
/. Environ. Eng. Sa. 2002, /, 17-31
25 Morris, J C, Isaac, R A In Water chlormation, 1983, Vol 4, pp 49-62
26 Trofe, T W, Inman Jr, G W, Johnson, J D, Kinetics of
monochloramme decomposition in the presence of bromide Environ. Sa.
Technol. 1980, 14, 544-549
27 Gallard, H, von Gunten, U, Chlormation of phenols Kinetics and
formation of chloroform Environ. Sa. Technol. 2002, 36, 884-890
28 Buffle, J Complexation reactions m aquatic systems: an analytical approach, Ellis
Horwood Limited Chichester, 1988.
29 Topudurti, K V, Haas, C N, THM formation by the transfer of active chlonne
from monochloramme to phloroacetophenone Jour.AWWA 1991, 83, 62-66
30 Isaac, R A, Morris, J C, Modeling of reactions between aqueous
chlorine and nitrogen compounds Water Chlormation: Environ. ImpactHealth Eff. 1983, 4, 63-75
31 Braun, W, Herron, J T, Kahanar, D K, ACUCHEM Computer program
for modeling complex reaction systems Int. J. Chem. Kjnet. 1988, 20, 51-62
32 Poskrebyshev, G A, Huie, R E, Neta, P, Radiolytic reactions of
monochloramine in aqueous solutions J. Phys. Chem 2003, 107, 7423-7428
33 Johnson, H D, Cooper, W J, Mezyk, S P, Bartels, D M, Free radical
reactions of monochloramine and hydroxylamine in aqueous solution
Radiât. Phys. Chem. 2002, 65, ^11-^,26
34 EECD Council Directive 98/83/EC of November 3 1998 on the qualityof water intended for human consumption, Official J L 330, 05/12/1998
35 USEPA National Primary Drinking Water Regulation Disinfection and
Disinfection Byproducts Final Rule, Federal Register, Vol 63, 241, 69389
36 Liang, L, Singer, P C, Factors influencing the formation and relative
distribution of halo acetic acids and trihalomethanes in drinking water
Environ. Sa. Technol. 2003, 37
37 Yao, J-J, Elovitz, M S, Miltner, R J Session 3, Group of IOA,
International Ozone Association, Florida, US Florida
38 Pedersen III, E J, Urbansky, E T, Marinas, B J, Margerum, D W,
Formation of cyanogen choride from the reaction of monochloramine
with formaldehyde Environ. Sa. Technol. 1999, 33, 4239-4249
39 WHO Guidelinefor drinking water quality, Volume I, Geneva, Switzerland, 1993
40 Obolensky, A , Philadelphia water district Personnal communication, 2003
6 Enhanced Bromate Control duringOzonation: Pre-oxidation with CICv
Marc-Olivier Buffle, Sonja Galli, Urs von Gunten,
O^one Science andEngineering, 2006
6.1 Abstract
Recently a number of control strategies —pH decrease, NH3 addiâon, Cl2-NH3
process— have been developed to minimize bromate formation during subsequent
ozonation. Here, we investigate the use of chlorine àoxide as a pre-oxidation step to
minimize bromate formation. The efficacy of the method depends on the type and
concentration of the natural organic matter (NOM) contained in the water. In water
from a mesotrophic lake containing 1.4 mgC/L DOC and 100 jUg/L Br,
pre-oxidation with 1 mgJT C102' àd not decrease bromate formation significantly.
However, when adding 1 mgC/L fulvic aad, more than 50% decrease in bromate
formation was observed. In waterfrom an eutrophic lake containing 3.2 mgC/T DOC
and 100 jUg/L Br, pre-treatment with 1 mg/L C102' decreased bromateformation by
more than 60"/o. In these experiments, the decrease in bromateformation coincided with a
decrease in HO'generated during ozpne decomposition. WhenpH was loweredfrompH
8 to pH 6 prior to ClOf pre-treatment, bromate formation was decreased 30 fold
from 25 to 0.85 jUg/L), however a decrease in HO' exposures was not observed,
inàcatmg that otherpH dependant mechanisms takeplace. Reactions of CIO2/CIO'
with many intermeàates bromine speaes (Bf, Br/, Br/, OBr, BrO') are not well
understood and are critical to fully understand the process. The addiâon of chlorite
(CIO/: the product of the reduction of C102' by NOM) prior to ozonation was also
able to minimize bromateformation although to a lesser degree than C102. This is likely
due to thefact that it is oxiàzed to C102 upon an electron transferfrom C102 to 03
before it isfully oxidized to chlorate, C103. From a kinetics standpoint, pre-oxidation
with C102 induces a lag-phase in theformation of bromate indicaâng that the method
mostly impacts bromateformation during the initialphase of ozpne decomposition. It can
therefore be advantageously combined with NH3 adàtion orpH decrease, which mostly
affect bromateformation during the secondphase of ozpne decomposition.
142 Chapter 6
6.2 Introduction
Potentially carcinogenic bromate is formed dunng the ozonation of
bromide-containing waters. The current dnnkmg water standards for
bromate in the EU, USA and WHO is 10 ug/L. In most waters
however, bromide concentration is below 100 ug/L. Bromide to
bromate conversion is generally below 10%, so that the bromate
standard is not reached is most waters (1). Nevertheless, a simultaneous
push towards lower bromate water standards and new disinfection
targets for Cryptosporiàum parvum oocysts (which requires high ozone
exposures) is generating interest in processes able to minimize bromate
formation during ozonation.
6.2.1 Existing bromate control strategies.
pH depression. Lowering the pH minimizes bromate formation
significantly by reducing the amount of HO' generated dunng
ozonation. It also shifts the acid-base equilibnum of HOBröOBr
—a key intermediate in the pathway from Br to BrOj — towards
HOBr, which is not oxidized by O3 (Figure 6.1) (2).
NH3 adàtion. NH3 reacts rapidly with HOBr to form NH2Br (Figure
6.1). NH2Br eventually reacts with ozone, recycling Br and NO3, but
the rate is slow and under standard process conditions most of the Br
is masked as NH2Br (2,3).
Cl2-NH3process. This process consists m a short pre-chlonnation step
followed by NH3 addition pnor to ozonation (4). There are multiple
mechanisms involved m the process. The most obvious is the oxidation
of Br by HOCl to HOBr, which following NH3 addition is masked as
NH2Br. This mechanism is different than NH3 addition m that NH2Br is
generated before ozone is added, so that the initial bromate formation
pathway through HO' is inhibited. Another important mechanism is
The CI02* pre-oxidation process for bromate minimization 143
the halogenation of some moieties of the natural organic matter
(NOM) such as ammo groups. Amines are partly responsible for HO'
generation dunng the initial phase of ozonation (time in the seconds
range). Recent investigations show that the halogenation of secondary
amines hmders the generation of HO' (5). A reduced HO' exposure
(JHO'dt) at a given ozone exposure results m a lower bromate
formation. Other mechanisms such as the reaction of HOBr with
NOM and the scavenging of HO' by NH2CI also play a role in the
efficiency of the process (4).
The ammonia addition and the pH depression processes can lead to a 50%
reduction in bromate formation. Weaknesses might mclude insufficient
Br03 -minimization, prohibition of ammonia addition to dnnkmg
water m certain countnes, and prohibitive costs of pH adjustments in
waters with medium to high alkalinity
The Cl2-NH3 process is more efficient; bromate can be decreased 8 fold
compared to standard ozonation and up to 40 fold when combined
with a decrease m pH Weaknesses mclude mcreased process
complexity and generation of small concentrations of halogenated
orgamcs dunng the pre-chlonnation step (4).
It is unlikely that one strategy will fit all applications. A broad palette of
bromate control strategies is therefore crucial for the water industry.
The weaknesses of the processes mentioned above demonstrate the
necessity to continue the development of new alternative bromate
control strategies. In previous pilot-scales investigations, the pre-
oxidation of natural waters with CIO2* before ozonation was shown to
significantly reduce bromate formation (6,7). In this article, we descnbe
the complexity of the mechanisms involved, investigate this process at
the lab scale, and confirm results obtained dunng pilot expenments.
144 Chapter 6
6.2.2 Pre-oxidation with CICV
From an operation stand-point, the rationale for applying a
pre-oxidation step before ozonation as a bromate control strategy is the
stabilization of ozone. In the water mdustry, disinfection requirements
are quantified as CT values. In a full scale system, CT can be calculated
by multiplying the residence time m a contact chamber with the oxidant
residual at the outlet of the chamber. If ozone decomposition is fast, a
large initial ozone concentration needs be added to obtain the targeted
residual at the outlet. This means that for the same calculated CT value,
faster ozone decompositions leads to larger net ozone exposures
(J03<it). The oxidation of a chemical compound is directly related to
the net oxidant exposure; hence, bromate formation in a full scale
system is reduced if ozone decomposition is slowed down after
installing a pre-oxidation step.
From a mechanistic stand-point, the rationale for the use of CIO2' as
a pre-oxidation step is the fact that CIO2* reacts very rapidly with
phenolic groups (8), which are ubiquitous constituents of the NOM.
Phenolic compounds generate high yield of HO' when ozonated (5)
and HO' have a key impact on bromate formation during the initial
phase of ozone decomposition (4,9). Hence, pre-oxidation of
phenolic moieties with CIO2* could decrease HO' generation and
therefore bromate formation. It should be noted that this is a
different mechanism than pre-oxidation with HOCl (m the Cl2-NH3
process). The reaction of phenolic moieties with CIO2* involve an
electron transfer, while the reaction with HOCl generates Cl-
substituted phenols which still generate HO' upon reaction with
ozone (5). In the use of HOCl as a pre-oxidation step, it is the
chlonne addition to amine moieties in the NOM that decreases the
generation of HO' upon ozonation (5).
The CI02* pre-oxidation process for bromate minimization 145
In reality, however, mechanisms in the C102*-based bromate
minimization process may be significantly more complex than what
is descnbed above. The top half of Figure 6.1 shows known
reactions of CIO2* and CIO2 with ozone and hydroxyl radicals (9).
Upon reaction with NOM, CIO2* is reduced to CIO2. If the
residence time between CIO2* addition and O3 addition is long
enough, CIO2* reacts entirely with NOM and ozone only encounters
CIO2 •The rapid reaction of ozone with CIO2 regenerates CIO2*
(see Table 6.1 for rate constants, ti/2 = 2 ms with 50 uM O3). It is
one of the rare cases of an electron transfer from an inorganic
species to ozone, i.e. an O3' is generated that instantaneously
decays to HO'. The reaction of HO' with CIO2 also leads to CIO2*.
CIO2* is then either oxidized to CIO3 by HO' or to CIO3' by ozone
(ti/2 = 13 s with 50 uM O3), which then reacts with CIO2* to form
CIO3 •Given the fast reaction of ozone with chlorite and relatively
slow reaction with CIO2', the ozonation of chlorite leads to a
significant simultaneous concentration of CIO2* during the initial
phase of ozonation. The presence of a CIO2* residual in the case
where it is not entirely reduced by NOM before ozone is added
does not fundamentally change the overall mechanism.
Hence, the full bromate minimization mechanism through addition
of CIO2* before ozonation is likely to involve more than the mere
pre-oxidation of phenolic moieties in the NOM, it probably also
includes direct interactions between C102*-denved species and
bromine species. When trying to tie the CIO2* and CIO2 oxidation
mechanisms with the bromate formation mechanisms, one
immediately notices that a full mechanistic understanding is not
possible due to a lack of knowledge on critical reactions (Table 6.1).
C102' and C102 do not react with Br (Table 6.1) and 90% of C102'
(hence >> 90% CIO2) is already oxidized 30 seconds after ozone
146 Chapter 6
addition (with 50 uM O3 and 6.6 picoM HO', data from Figure 6.4).
It is therefore unlikely that reactions between C102*-denved species
and bromine species would take place dunng the second phase of
ozonation (minute range). Dunng the initial phase of ozonation, the
HO' oxidation pathway involving Br', Br2*, Br3*, BrO', OBr and
Br02 is key. The reactions between these bromine species and CIO2*
or CIO2 however, are not known.
CIO2' -^ C102- j—^ CI02-
o3 o3
\
C103
Br
H2>HOBr/IM' I U
> OBr -^BrOr
nh3
NH2Br
Br03-
Figure 6.1. Top mechanisms of CIO2" and CIO2 reactions with O3
and HO" Bottom bromate formation mechanism), adapted from
von Gunten (2003)
The CI02* pre-oxidation process for bromate minimization 147
Finally, the scavenging of HO' by CIO2* and CIO2 might be another
mechanism of importance. In Lake Zunch water, the scavenging rate
of HO' mcreases by 20-60% due to the addition of the chlonne species
(with 0.5-1 mg/L CIO2* or CIO2 and constant from Table 6.1),
however, given the generation of O3' upon O3 reaction with CIO2,
prediction of the net HO' yield is not straightforward.
Table 6.1. Second-order rate constants at 20 °C
k" [M is 1] Ref
C102 + O3 -> CIO/ + HO- 8 2 x 10« (10)
CIO2 + HO- -> C102- 6 3 x 109 (11)
C102- + O3 -> C103- 1 05 x 103 (12)
QO2 + HO- -> C1Q3 4xl09 (12)
C102- + Br -> product <<104 (13)
C102- + Br2- -> product 1 2 x 109 (14)
a02 /Q02/QQ3-+Br-/ Br2VBr3- /BrO-/OBr/Br02->N/A N/A
6.2.3 Previous large-scale investigations on bromate
control with CICV
Expenments conducted at a full-scale demonstration plant
(2.2-3.1 mgC/L TOC, 100-500 ug/L Br) demonstrated that an addition
of 1 mg/L CIO2' reduced bromate formation significantly
(50-90% depending on the ozone exposure) (6). The data shows slow
bromate formation at small CTs (lag-phase) followed by a more rapid
formation at larger CTs. For the same 03-doses, CTs were larger after
pre-oxidation with CIO2', indicating that the stabilization of ozone due to
pre-oxidation was more important than the scavenging effect of the
formed chlonte on ozone. Bromate formation was also minimized when
CIO2 was added. However, for the same ozone dose, no gam m
disinfection could be observed compared to standard ozonation.
pH depression combmed with CIO2' pre-oxidation decreased bromate
formation further.
148 Chapter 6
6.3 Materials and Methods
6.3.1 Reagents
All reagents were of analytical grade. All solutions were prepared with
MilliQ water with a resistivity larger than 18 MQ-cm. Aqueous ozone
was prepared as descnbed elsewhere (4), the ozone stock solution
concentration was typically 1.6 mM. Stock solutions of CIO2' were
prepared as m (15). The fulvic acid extract (Reference FA, Nordic Lake)
was obtained from the International Humic Substance Society.
6.3.2 Natural Waters
Lake Zunch water (pH 7.8, alkalinity 2.4 mM, 5 ug/L NH3,
DOC 1.4 mgC/L) was collected from the raw water mtake of the Lengg
dnnkmg water treatment plant, 30 meters below the lake's surface, while
lake Greifensee water (pH 8.2, alkalinity 3.4 mM, 370 ug/L NH3,
DOC 3.4 mgC/L) was collected at the natural outlet of the lake. Waters
from Lake Zunch (mesotrophic) and lake Greifensee (eutrophic) were
filtered at 0.45 urn (cellulose nitrate filters, Sartonus) and kept at 4°C.
6.3.3 Methods and procedures
The expenmental procedure is displayed m Figure 6.2. The water was
buffered with 2 mM borate for pH 8 (phosphate for pH 6 and 7), spiked
with pCBA (1 uM), bromide (100 ug/L) and fulvic acid (1 mgC/L).
CIO2' (0.5 or 1 mg/L) was then added and left to react 10 minutes with the
natural water. The water was then transferred to an open beaker and stirred
vigorously dunng 20 minutes to remove any CIO2' residual (> 80%
degassed after 5 minutes of stirring). A sample was taken to control
photometncally that no CIO2' residual was left m solution, following which
a blank was taken. Ozone was then added to the water and samples were
The CI02* pre-oxidation process for bromate minimization 149
taken at regular time mtervals. All kmetics expenments were performed at
20°C with 500 ml amber glass bottles capped with a dispenser. Three vials
containing reagents were filled for each time point (Figure 6.2). HO'
exposure was back-calculated usmg the oxidation of para-chlorobenzoic
acid (pCBA) analyzed with HPLC m the range of 0.05 to 1 (oM (16).
Ozone was measured with a Vanan CarylOO at 258 nm (e = 3000 M kma)
or with the mdigo reagent at 600 nm (e= 20'000 M kma) (17). Bromate was
measured with IC and UV detection after a post-column reaction; the
method has a quantification limit of 0.5 ug/L (18). CIO2' was measured
with ABTS dye, with a quantification limit of 10 ug/L (19).
03 photon w/ 200 400 uVi Indigo +0 15 M H3PO4
HO pCBAw/HPLC quenched w/1 25 mM thiosulfete
BrOs with C UV quenched»// 200 400 uVi Indigo
Figure 6.2. Expérimental procedure followed during this investigation
6.4 Results and Discussion
6.4.1 Impact of dose and pH on CIO2* decomposition
Figure 6.3a shows the decrease of CIO2' at 3 different doses 0.13, 0.25 and
0.5 mg/L m Lake Zunch water. At a dose of 0.5 mg/L, 40% of CIO2' reacts
m 10 minutes, while at 0.13 mg/L all CIO2' has reacted pnor to 10 minutes.
It can be assumed that roughly all C1CV that has reacted is reduced to CIO2.
150 Chapter 6
0 600 1200 1800 2400 3000 3600
Time [s]
o
ü
1200 1800 2400
Time [s]
Figure 6.3. C102" decrease as a function of time in Lake Zurich water
at various doses and pH (a) C102" doses of 0.13, 0.25 and 0.5 mg/L at
pH 8 (b) C102- dose of 0.2 mg/L at pH 6, 7 and 8.
The CI02* pre-oxidation process for bromate minimization 151
Figure 6.3b shows that an increase in pH accelerates CIO2*
decomposition significantly CIO2* exposure (JC102*dt) after 10 minutes
at pH 8 is 40% smaller than at pH 6. This behavior is expected based
on the much higher rate constants for the reactions of CIO2' with
deprotonated compounds (13). NOM contains amines and activated
aromatic compounds which upon deprotonation react more readily
with C102'.
6.4.2 Impact of CI02* pre-treatment on ozone decomposition
Little change was observed m the decomposition of ozone in non-
treated waters and m water pre-oxidized with CIO2' (data not shown).
This could be explained by the fact that although moieties pre-oxidized
with CIO2* might not react with ozone anymore, CIO2' is reduced to
CIO2 which then reacts with ozone, i.e. one observes a stoichiometnc
status quo. The similanty between ozone decomposition profiles
following vanous pre-treatment conditions would allow a direct
companson of bromate formation as a function of time. However, the
data is presented as a function of ozone exposure (JOjdt), so that a
companson across different waters and doses can be made.
6.4.3 Impact of water quality on Br03" minimization with CI02*
Figure 6.4 shows the impact of CIO2* pretreatment on HO'
generation and bromate formation. Take Zurich water containing
100 ug/L Br was pre-oxidized with 0.5 mg/L CIO2* for 10 minutes
before 1.5 mg/L ozone was added. For this particular water, the
CIO2* pre-oxidation did not have much impact (Figure 6.4a). At the
first measurement (1 mg/L-mm = 30 seconds), bromate formation
was decreased by roughly 30%, however there is no improvement at
higher ozone exposures. The difference in HO' generation is
insignificant at any given ozone exposure, indicating that the
generation of HO' upon ozonation of CIO2 compensates for the
152 Chapter 6
added HO'-scavengmg by CIO2* and CIO2 and the decrease of HO'
generated by ozone following the pre-oxidation of NOM. Lake
Zurich water contains very small concentration of NOM
(DOC = 1.4 mgC/L), so that it could be expected that the impact
of pre-oxidation of NOM to minimize bromate formation would
be small.
The same expenment was performed with Take Greifensee water
which contains a higher concentration of NOM (DOC= 3.4 mgC/L) but also a high natural concentration of
NH3/NH4+ (370 ug/L). The water was also spiked to contain
100 ug/L Br. The CIO2* and O3 doses were doubled according to
the higher DOC concentration. Pre-oxidation with CIO2* decreased
HO' generation but the difference was small (Figure 6.4b). The
formation of bromate, however, decreased roughly by a factor of 4
at 1 mg/L-mm (= 30 s) and a factor of 2 at 6 mg/L-mm (= 10 mm).
The formation of bromate after pre-oxidation with CIO2* (solid
squares) shows an inflexion (lag-phase) at lower exposures; this
feature was noticed throughout the investigation. This characteristic
lag-phase can also be observed in the data presented by
Krasner et al. (2004). It confirms that it is the initial phase of
bromate formation —controlled by the HO' pathway— that is
hindered (see introduction).
When comparing bromate formed in untreated Lake Zurich and
Lake Greifensee waters (open symbols), it is striking to note the
difference at the same ozone exposure (e.g. 18 vs. 9 ug/L Br03 at
5.5 mg/L-mm, respectively). This is due to the high natural
concentrations of NH3/NH4+ contained in Lake Greifensee water
(370 ug/L). A 2 fold reduction in bromate formed was also found
in previous research investigating the addition of NH3 to minimize
bromate formation (3,4). After pre-oxidation with CIO2', bromate
The CI02* pre-oxidation process for bromate minimization 153
formation in Lake Greifensee water was much smaller than in Lake
Zurich water. This is the result of a combination of two
complementary effects: (1) the pre-oxidation with CIO2* which is more
noticeable in Greifensee water due to its larger DOC concentration and
which affects bromate formation pnor to 30 seconds, and (11) the
presence of ammonia in Greifensee water which mostly affects the
second phase of bromate formation (mmute range).
Fulvic acids extract (1 mgC/L) was added to Lake Zurich water to
simulate a natural water containing a higher NOM concentration
while maintaining a low NH3 concentration (Figure 6.4c).
Compared to unmodified Lake Zunch water (Figure 6.4a), bromate
formation in water spiked with fulvic acids is increased significantly,
especially at the first data point (from 3.6 to 8.3 ug/L BrOs).
This increase can be well explained by the enhanced generation of
HO' pnor to the first measurement at 30 seconds in the fulvic
spiked water (JHO'dt increases 3 fold from 6.3 x 10 u M-s to
20 x 10 n M-s). The strong impact of fulvic and humic acids on the
initial generation of HO' was reported in a previous study (5).
The effect of pre-oxidation dose (0, 0.5 and 1 mg/L CIO2') on the
fulvic acid containing Lake Zunch water shows a clear trend
towards lower HO' exposures and lower bromate formation as
CIO2* dose is increased. However, it seems that for 1 mg/L CIO2',
bromate minimization at 30 seconds (from 8.3 to 1.4 ug/L Br03 )
much surpasses that of HO' exposure decrease (JHO'dt decreases
from 20 x 10 u M-s to 12 x 10 n M-s), supporting the hypothesis
that additional mechanisms are of importance.
154 Chapter 6
Figure 6.4. HO" generation and bromate formation following C102" pre-
oxidaüon and subsequent ozonation of various -waters spiked -with
100ug/L Br (a) Lake Zurich water (DOC=l 4 mg/L), pH 8, 0/0 5mg/LC102', 1 5 mg/L O3 — (b) Lake Greifensee water (DOC 3 2 mg/L),pH 8, 0/1 mg/L C102", 3 mg/L O3 — (c) Lake Zurich -water spiked with
lmgC/L fulvic acid, pH 8, 0/0 5/1 mg/L C102-, 3 mg/L O3
The CI02* pre-oxidation process for bromate minimization 155
6.4.4 Impact of pH on bromate minimization with CI02*
Lake Zunch water spiked with 1 mgC/L fulvic acids and pre-treated
under vanous conditions was ozonated with 3 mgOs/L at pH 6, pH 7
and pH 8 and bromate concentrations were measured. In non-treated
Lake Zunch water (Figure 6.5a), the difference in bromate formed at
pH 8 and at pH 6 (-30%), is not as large as the 50% observed
previously (4). This is due to the fact that this water being spiked with
fulvic acids, it contains a larger fraction of phenolic compounds than
the water tested in Buffle et al. (2004). Phenolic moieties are in large
part responsible for the generation of HO' and they react so rapidly
with ozone that a decrease from pH 8 to pH 6 does not noticeably
decrease HO' generation (5). Hence, decreasing the pH does not
minimize bromate formation during the initial phase of ozonation m
this water. In fact, if the bromate concentration is corrected for its
initial formation, a 50% decrease can mdeed be observed between
pH 8 and pH 6. Similarly to NH3 addition, pH mostly affects the
second phase of bromate formation.
In Lake Zunch water, combining a pre-treatment with 1 mg/L CIO2*
and a decrease in pH has a synergistic effect on bromate
minimization. In Figure 6.5b, at an ozone exposure of 6 mg/L-mm
(E99% inactivation of Cryptosporidium parvum at 20°C (20)) bromate
is decreased by a factor of 18 (from 15 to 0.85 ug/L BrOs) for a pH
reduction from pH 8 to pH 6. It should be noted that at pH 6 and
pH 7, bromate is below the quantification limit (0.5 ug/L) up to an
ozone exposure of 2 mg/L-mm (> 2 logs inactivation of Giardia
Muns cysts).
156 Chapter 6
L Zurich water L Zurich water + 1 mg/L CI02'
b)
ozone exposure [mg/L mirV0 2 4 6
ozone exposure [mg/L mm]
Figure 6.5. Bromate formation following ozonation (3 mg/L O3) of
Lake Zurich -water spiked -with 1 mgC/L fulvic acid and 100 ug/L Br
at pH 6, 7, 8 and various pretreatment conditions (a) untreated -water
(b) -water pre-oxidized -with 1 mg/L C102" (c) -water spiked -with
0 5 mg/L C102 (d) water pre-oxidized with 1 mg/L C102" but not
degassed after 10 minutes The horizontal dashed lines represent the
bromate drinking -water standards of 10 ug/L
Lake Zunch water was spiked with 0.5 mg/L CIO2, which corresponds to
the concentration of reduced CIO2', 10 mmutes after addition of 1 mg/L
CIO2'. A significant decrease m bromate formation could also be observed
as pH is decreased (Figure 6.5c). The pattern of the bromate formation
curves are similar to that of CIO2' pretreatment, with the lag-phase m
The CI02* pre-oxidation process for bromate minimization 157
bromate formation mcreasmg at lower pH This is a strong mdication that
the bromate minimization mechanism is similar for CIO2 and for CIO2'.
Figure 6.5d shows bromate formation m water pre-treated with 1 mg/L
CIO2', however, CIO2' was not degassed (i.e. the solution still contains a
CIO2' residual). The results ke somewhere m between the results of the
preceding two experiments, suggesting that no crucial additional mechanism
take place. In all cases descnbed above, there was no clear correlation
between decrease m bromate formation and HO' generation. This confirms
the existence of an alternative bromate minimization mechanism, potentially
mvolvmg reactions between bromine speaes and Q02*-denved species.
6.5 Conclusions
The use of chlonne dioxide as a pre-treatment step pnor to ozonation to
minimize bromate formation can be very effiaent but is highly dependant
on the type and concentration of NOM contained m the water. The above
expenments clearly determined that chlonne dioxide strongly disrupt the
initial phase of bromate formation. It is therefore a method of choice to
combme with control stratégies that are mostly effiaent m the second
phase of bromate formation such as NH3 addition and pH reduction.
Experiments m waters where ammonia was present or pH was depressed
showed up to a 30 fold decrease m bromate formed. The use of chlonne
dioxide as a pre-treatment step has an advantage over the CI2-NH3 method
m that no tnhalomethane or other haloorganics are formed. However, the
ozonation of chlonne dioxide residual or its reaction product with NOM
—chlonte— leads to formation of chlorate m nearly 100% yield
(i.e. 1.24 mg/L chlorate per mg/L CIO2'). An action level for chlorate of
0.8 mg/L has been set m the state of California (6) and a tolerance value of
0.2 mg/L m Switzerland, so that the appkcation of CIO2' automatically
impkes an optimization procedure. A complete understanding of the
mechanisms mvolved dunng the process is still lacking, further research is
therefore warranted.
158 Chapter 6
6.6 Acknowledgments
We thank CIRSEE - Suez Environnement for financial support; Isabelle
Baudm, Auguste Bruchet, Zdravka Do-Quang, Mane-Laure Janex, Jean-
Michel Lamé (CIRSEE) for fruitful discussions; Michael Dodd, Marc
Huber, Gretchen Onstad and Elisabeth Salin for insightful comments.
6.7 References
1 von Gunten, U, Salin, E, Bromate in drinking -water a problem in
Switzerland? O^one: Science & Engineering 2003, 25, 159-166
2 von Gunten, U, Hoigné, J, Bromate formation during ozonation of
bromide-containing -waters lnterraction of ozone and hydroxyl radical
reactions Environ. Sa. Technol. 1994, 28, 1234-1242
3 Pinkernell, U, von Gunten, U, Bromate minimization during ozonation
Mechanistic considerations Environ. Sa. Technol. 2001, 35, 2525-2531
4 Buffle, M-O, Galli, S, von Gunten, U, Enhanced Bromate Control
during Ozonation The Chlorine-Ammoma Process Environ. Sa. Technol.
2004, 38, 5187-5195
5 Buffle, M-O, von Gunten, U, Phenol and Amine-lnduced HO'
Generation During the Initial Phase of Natural Water Ozonation
Environ. Sa. Technol, accepted, 2006
6 Krasner, S W, Yun, T, Yates, R, Mofidi, A, Liang, S, Coffey, B Pre-
oxidation -with chlorine dioxide to control bromate formation during
subsequent ozonation San Antonio, Texas
7 Neemann, J Use of Chlorine Dioxide and O^onefor Control of Disinfection By¬products, AwwaRF Report, 2005
8 Tratnyek, P G, Hoigné, J, Kinetics of reactions of chlorine dioxide
(OCIO) in -water - II Quantitative structure-activity relationships for
phenokc compounds Wat. Res. 1993, 28, 57-66
9 von Gunten, U, Ozonation of drinking -water Part II Disinfection and
by-product formation in presence of bromide, iodide and chlorine Wat.
Res. 2003, 37, 1469-1487
10 Nico son, J S, Wang, L, Becker, R H,Huff Hartz, K E
, Müller, C E,
Margerum, D W, Kinetics and mechanisms of the ozone/bromite and
ozone/chlorite reactions Inorg. Chem. 2002,41, 2975-2980
The CIO2" pre-oxidation process for bromate minimization 159
11 Buxton, G V, Subhani, M S, Radiation chemistry and photochemistryof oxychlorine ions 1 Radiolysis of aqueous solutions of hypochloriteand chlorite ions Journal of the Chemical Society, Faraday Transactions 1:
Physical Chemistry in Condensed Phases 1972, 68, 947-957
12 Klärung, U K, Sehested, K, Holcman, J, Standard Gibbs energy of
formation of the hydroxyl radical in aqueous solution Rate constants for
the reaction chlorite (C102-) + ozone dblarw ozone(l-) + chlorine
dioxide Journal of Physical Chemistry 1985, 89, 760-763
13 Hoigaé, J, Bader, H, Kinetics of reactions of chlonne dioxide (OQO) in water -
I Rate constants for inorganic and organic compounds Wat. Rés. 1994, 28, 45-55
14 Mialocq, J C, Barat, F, GUles, L, Hickel, B, Lestgne, B, Flash photolysis of
chlonne dioxide in aqueous solution Journalof PhysicalChemistry 1973, 77, 742-749
15 Huber, M M, Korhonen, S
, Ternes, T A,von Gunten, U, Oxidation of
pharmaceuticals during water treatment with chlorine dioxide Water
Research 2005, 39, 3607-3617
16 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios duringozonation processes I The Rct concept O^one Sei. Eng. 1999, 21, 239-260
17 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Wat. Res. 1981, 15, 449-456
18 Salhi, E, von Gunten, U, Simultaneous determination of bromide,
bromate and nitrite in low ug 1-1 levels by ion chromatography without
sample pretreatment Wat. Res. 1999, 33, 3239-3244
19 Pinkernell, U, Nowack, B, Gallard, H ,von Gunten, U, Methods for the
photometric determination of reactive bromine and chlorine species with
ABTS Wat. Res. 2000, 34, 4343-4350
20 Finch, G R, Belosevic, M, Controlling Giardia spp and
Cryptosporidium spp in dunking water by microbial reduction processes
/. Environ. Eng. Sei. 2002, /, 17-31
AI Oxidation of Antibacterial Molecules by
Aqueous Ozone: Moiety-Specific Reaction
Kinetics and Application to Ozone-Based
Wastewater Treatment
Michael C Dodd, Marc-Olivier Buffle, Urs von Gunten
Environmental Science and Technology, 2006
1.1 Abstract
Ozpne and hydroxyl radical ('OH) reaction kinetics were measured for 14
antibactenal compounds from nine structural families, to determine whether
muniapal wastewater ozonation is likely to result in selective oxidation of these
compounds' biochemically essential moieties These compounds are oxidized by ozpne
with apparent second-order rate constants, k03app
> 1 x 103 M's', atpH 7, with
the exception of N(4)-acetylsulfamethoxazple (k03app is 25 x 102 M's')
k03 spp @>H 7) for macrolides, sulfamethoxazole, tnmethopnm, tetracycline,
vancomycin, and amikacin appear to correspond directly to oxidation of
biochemically essential moieties Initial reactions of ozpne with
N(4)-acetylsulfamethoxazple, fluoroquinolones, lincomyan, and ß-lactams do not
lead to appreaable oxidation of biochemically essential moieties However, ozpne
oxidizes these moieties within fluoroquinolone and lincomyan via slower reactions
Measured k03app values and second-order 'OH rate constants, k,0Happ,
were
utilized to charactenze pollutant losses dunng ozonation of secondary muniapal
wastewater effluent Measured losses were dependent on k03 app,
but independent of
k.oh app Ozpne doses ~>3 mgJT yielded ~>99"/o depletion of fast-reacting
substrates (k03apP > 5 x 104 M's') at pH 7 7 Ten substrates reacted
predominantly with ozpne, only four were oxidized predominantly by 'OH These
results indicate that many antibactenal compounds will be oxidized in wastewater
via moiety-speafic reactions with ozpne
162 Appendix I
1.2 Introduction
ClnncaUy-important antibacterial agents are virtually ubiquitous
contaminants of municipal wastewaters (Supporting Information, Figure
SI) (1-5). Raw and primary wastewaters typically contain compounds
from various antibacterial classes at individual concentrations ranging
from 0.5-3 ug/L (Figure SI). These concentrations can often be reduced
by 60-90% during conventional activated sludge treatment combined
with tertiary filtration (2,4,6). This is generally sufficient to achieve
effluent concentrations below levels known to be detrimental to bacteria
(Figure SI) and other aquatic life (7,8). However, effluent concentrations
of certain antibactenals (e.g., fluoroquinolones (2,3)) may still be harmful
to organisms present in effluent-dominated receiving waters (9). In
addition, because activated sludge processes typically operate at solids
retention times of several days, conventional wastewater treatment results
in prolonged exposure of wastewater-borne bacteria to significantly
higher antibacterial concentrations than are present in treated effluents
(2,4-6). In the case of fluoroquinolones, these concentrations can
approach minimal growth inhibitory concentrations (MICs) for TL. coll
(Figure SI) and other bacterial strains (10) - a condition which may favor
evolution of low-level antibacterial resistance m affected bacterial
communities (11,12).
In the interest of precaution, unnecessary exposure of wastewater-
borne and environmental microbiota to biochemical stress originating
from antibacterial compounds should be minimized when possible.
Supplemental wastewater treatment technologies capable of yielding
rapid biochemical deactivation of such compounds could aid m
achieving this ob|ective. Ozonation, which is utilized in advanced
treatment of secondary wastewater effluent (13), and has shown
potential as a means of pre-oxidizmg and disinfecting primary
wastewater effluents (14,15), appears promising m this regard (16-18).
Moiety-specific oxidation of antibacterial molecules 163
Recent studies have shown that relatively low ozone (O3) doses (< 5
mg/L) can yield > 90% depletion of many antibactenal compounds in
wastewaters containing up to 23 mg-C/L of DOC (18,19). Although
mineralization of antibactenal molecules will be mfeasible dunng
municipal wastewater ozonation, partial oxidation may be sufficient to
achieve their biochemical deactivation, provided that O3 reacts with the
parent molecules in a manner leading to rapid, selective oxidation of
functional moieties related to their antibactenal activities. Such an
outcome has been demonstrated for the steroid hormone 17a-
ethmylestradiol, in which case ozone selectively oxidizes the phenol
moiety responsible for the parent molecule's estrogenic activity (20).
Similar results appear likely for many antibactenal molecules (Table 1,
Text SI). However, some antibactenal compounds' biochemically-
essential moieties may be 03-refractory, or O3 may react preferentially
with moieties nonessential to the parent molecules' biochemical
activities (Table 1). In such cases, relative reactivities of "essential" and
"nonessential" functional moieties will influence the likelihood that
antibactenal compounds can be biochemically deactivated dunng
wastewater ozonation.
pH-dependent vanations in measured "apparent" rate constants can be
used to determine "specific" rate constants for reactions of O3 with
each individual acid-base species of an îonizable substrate (21-23). This
approach was applied in the present study to evaluate O3 reaction
kinetics measured for fourteen antibactenal molecules representing
nine of the most widely-used antibactenal structural families (Table 1).
O3 reaction kinetics for substructure model substrates (Table 2)
representing the theonzed reactive moieties withm each antibactenal
molecule were measured to facilitate assignment of calculated species-
specific reactivities to individual functional moieties.
164 Appendix I
Pollutant transformation dunng wastewater ozonation is also
influenced by hydroxyl radicals (*OH) generated through reactions of
O3 with specific functional moieties in dissolved organic matter (24,25)
or from auto-catalytic O3 decomposition (26). Because »OH reacts
rapidly with a wider vanety of functional moieties than O3, the
oxidative specificity of an ozonation process may be diminished if
dominated by »OH reactions. The importance of »OH and O3 in the
context of antibactenal compound oxidation was investigated by using
rate constants determined here to charactenze observed antibactenal
compound losses dunng ozonation of a secondary wastewater effluent.
Moiety-specific oxidation of antibacterial molecules 165
Table 1. Antibacterial substrates and expected sites of O3 attacka>b>c>d
Macrolides
\2.
°3
Roxithromycin (RX)
pKa = 9 2
Azithromycin (AZ)
pKah2 = 8 7, 9 5
^i^b^oiTylosin (TYL)
pKa = 7 7
Sulfonamides
'"*. 2N-<
^1
Sulfamethoxazole (SMX)
pg,1-2=17,5 6
A^(4)-acetyl-sulfamethoxazole
(ASMX)
pKa = 5 5
DHFR Inhibitor"
Trimethoprim (TMP)
piTal,2=3 2,7 1
Fluoroquinolones
00 ö ö -
Ciprofloxacin (CF) Enrofloxacm (EF)
pKah2 = 6 2, 8 8 piTal,2=6 1,7 7
Lincosamide
OH \
HOiYHN A
Lincomycin (LM)
pKa = 7 8
ß-lactams Tetracycline
Penicillin G (PG)
pg, = 2 7
Cephalexin (CP)
pg,1-2 = 2 5, 7 1
Tetracycline (TET)
yK,w = 3 3, 7 7, 9 7
Glycopeptide
HCYY
Aminoglycoside
Amikacin (AM)
P^ai,2,3,4 = 6 7, 8 4, 8 4, 9 7
Vancomycin (VM)
P^al,2,3,4,5,6 =
2 9, 7 2, 8 6, 9 6, 10 5, 11 7
^Structural families are listed in bold. bO?> target sites are classified as: "essential"
(i.e., these moieties are directly responsible for the parent molecules' antibacterial
activities — Text SI) — indicated by solid arrows, and "nonessential" (i.e.,thesemoieties are not directly responsible for the parent molecules' antibacterial
activities) — indicated by dotted arrows. "References from which pKa values were
obtained are summarized in Supporting Information, Table SI. ^Sites of ionization
are numbered according to order of deprotonation. For compounds possessing a
single pKs., the îonizable site is labeled 1. ÜHFR — dihydrofolate reductase.
166 Appendix I
Table 2. Substructure model substrates and expected sites of O3 attack*1 h
£0. X°30
°3 Ko_y
A^A^-dimethylcyclo-
hexylamme (DMCH)
1 -methylpyrrohdme
(MP)
4-ammophenyl methylsulfone (APMS) 3,5-dimethyhsoxazole
(DMI)pK, = 10 7 pK,= \0 2 pK,= \5
Model for RX, AZ,
TYL, TETModel for AZ, LM Model for SMX
Model for SMX,
ASMX
öA—1o/^\°^
0
fXXUoh
_^°3
£rEthyl A^-piperazme-
carboxylate (EPC)
pKa = S3
Flum equine (FLU)
pK, = 6 5
2,4-diammo-5-
methylpyrimldme
(DAMP)
pK,h2=3 2,ll
3,4,5-
trim ethoxytoluene
(TMT)
Model for CF, EF Model for CF, EF Model for TMP Model for TMP, VM
Vf?0 OH
Jmh2
°"6 c^C2-(3-methylbutyryl)-5,5-
dimethyl-1,3-
cyclohexandione
(MBDCH)
pK, = 3 5
Cyclohexylamme
(CH)
p£,= 10 6
Cyclohexane-
methylamme (CHM)
pK,= \0 3
Model for TET Model for AM Model for AM
^References from which pK& values were obtained are summarized in SupportingInformation, Table S2. b In the case of DAMP, which has two pK^s, sites of
ionization are numbered according to order of deprotonation. Sites of ionization
are labeled 1 for all other îonizable compounds.
Moiety-specific oxidation of antibacterial molecules 167
1.3 Materials and Methods
1.3.1 Chemical Reactants and Reagents
All reagents and reactants were of 95% purity or greater, with the exception
of ASMX, which was ~70% pure. Descriptions of chemical sources and
stock solutions are provided in Supporting Information, Text S2.
1.3.2 Measurement of Rate Constants
O3 and «OH rate constant measurements were conducted according to
seven expenmental protocols (designated I to VII), which are
summarized m Table 3 and described m detail withm Text S3. Solution
pH was maintained in all kinetic expenments with phosphate buffers
of approximately 10-mM concentration. Ten-mM r-BuOH was added
as a «OH scavenger to solutions used for measurement of O3 rate
constants. O3 and «OH rate constants were measured at 20(+0.5) °C
(in accordance with previously determined O3 reaction
kinetics (17,21,27,28)) and 25(±0.5) °C, respectively.
Table 3. Experimental methods used for rate constant measurements
(details in Text S3)
Method Oxidant Experimental Procedure" Measurement Endpoint
03 loss (measured at X = 258 nm)
Substrate loss (measured by HPLC)
Reaction product yields (measured by various
techniques )
Losses of each competitor (measured by HPLC)
Losses of each competitor (measured by HPLC)
Losses of each competitor (measured by HPLC)
Losses of each competitor (measured by HPLC)
^SFL stopped flow spectrophotometry, CK — competition kinetics, ^Reaction
products were measured either spectrophotometncally or by HPLC, dependingon the analyte
I 03 SFL
II 03 Batch
III 03 CK
IV 03 CK
V •OH CK (H202-photolysis)
VI •OH CK (y-radiolysis)
VII •OHCK (H202-photolysis, amine
denvatization)
168 Appendix I
1.3.3 Wastewater Matrix Experiments
Wastewater expenments were conducted in batch, by momtonng loss
of each substrate in a sample of Kloten-Opfikon secondary wastewater
effluent for vanous O3 doses, at 20(+0.5) °C (see Text S4 for
expenmental details). Sample pH, alkalinity, and DOC were 7.7,
3.5 mM as HCO3 and 5.3 mg-C/L, respectively.
1.4 Results and Discussion
1.4.1 Moiety-specific Ozone Reaction Kinetics
pH-dependent, apparent second-order rate constants, k"o3,aPP, were
determined for each substrate at pH values ranging from 3 to 8,
according to the methods descnbed within the Supporting Information
(Text S3). pH-dependencies of measured k"o3,aPP values were modeled
according to a modified second-order rate expression (eq 1) that
incorporates acid-base speciation of substrate, M,
Ä=iö=_, [o3lM]T=-^,[03]a[M]T W
dt rj dt,=1
where [M]t represents the total concentration of M (including all n acid-
base species), r\ represents an apparent stoichiometnc factor accounting for
moles of O3 consumed per mole of substrate consumed, k^ is the specific
rate constant corresponding to reaction of O3 with substrate acid-base
species 1, and a, represents the equilibnum distnbution coefficients for
species 1. k't values — summanzed in Tables 4 and S3 — were calculated by
nonlinear regression of expenmental data according to eq 2,
n
kO„app,M =T,k"a> (2)
via a Marquardt-Levenberg curve-fitting routine (SigmaPlot 2002, SPSS
software).
Moiety-specific oxidation of antibacterial molecules 169
1.4.1.1 Macrolides.
Roxithromycin (RX). The strong pH-dependency of k"03jilpBRx
(Figure la) indicates that O3 reacts initially at the RX structure's neutral
tertiary amine moiety (Table 1). The continuous decrease in k"o3,aPBRX
with decreasing pH — due to protonation of RX's tertiary amine
(17,21,22) — suggests that O3 reactivity with the remainder of the RX
structure is very low, and that oxidation at circumneutral pH will occur
exclusively at the deprotonated tertiary amine. The pH-dependency and
magnitudes of k"o3,aPP measured for N,N-dimethylcyclohexylamine
(DMCH, Table 2) (Figure la) support these conclusions.
Azithromycin (AZ). k"o3,apP,AZ exhibits nearly the same pH-
dependency as k"03japBRx (Figure la). However, the close proximity of
pKai and pK^ values for AZ prevents one fiom determining whether
k"o3,aPBAZ is due pnmanly to reaction of O3 with the parent molecule's
exocyclic tertiary amine or with its heterocyclic tertiary amine (Table 1).
Companson of the rate constant for reaction of neutral
1-methylpyrrolidme (MP in Table 2) with O3 to that for neutral DMCH
shows that the two values are quite similar (Figure la). This suggests
that the corresponding moieties in the AZ structure (Table 1) react
with O3 at roughly equivalent rates.
Tylosin (TYL). TYL reacts with O3 significantly faster than RX and
AZ at acidic pH (Figure la). This can be attnbuted pnmanly to the
con|ugated diene moiety withm TYL's macrolactone ring (Table 1).
Olefins typically react with O3 at rates that are independent of pH,
within the range 105-106M h 1, unless substituted with strong electron-
withdrawing groups (27,31). The rate constant reported for the neutral
form of the model diene sorbic acid (k'03 = 3.2 x 105 M h 1 (32)) is
within a factor of ~4 of that for the TYL cation (Figure la). This
provides additional evidence that the diene moiety is responsible for
TYL's high reactivity toward O3 under acidic conditions. At pH > 6, the
170 Appendix I
proportion of neutral TYL (i.e., deprotonated tertiary amine) becomes
high enough to influence the magnitude of k"o3^ppi,TYL, and dominates
measured reactivities for more alkaline pH ranges (Figure la).
Table 4. Second-order rate constants (M 1s ^ for reactions of
antibacterial substrates with O3 and •OH'3
Substrate6
(Rate constant
measurement
methods')diprotonated
speaes
monoprotonated
species
deprotonated
species
ft03,app
(PH7)
b-"ft03,app
»H 7.7)
^•OH,app'
(PH7)
RX (I, V) NA7 <l(17f 10 (±0 l)x 101 (17)k 6 3 x 104 31 xlO5 5 4 (±0 3)xlOs
AZ (I, V) <1' 6 0 (±1 l)xl06 6 0 (±219)xl06' 1 1 x 105 52 xlO5 2 9 (±0 6)xlOs
TYL (IV, VI) NA7 7 7 (±14)xl04 2 7 (± 0 5) x 106 5 1 x 105 14 xlO6 8 2 (±0 1)xl0s
SMX (TV) nl/ 4 7 (±0 9)xl04 5 7 (± 1 0) x 105 5 5 x 105 57 xlO5 5 5 (± 0 7) x 10s (17)
ASMX (H, V) NA7 2 0 (±0 2)xl0' 2 6 (± 0 1) x 102 2 5 x 102 26 xlO2 6 8 (± 0 1) x 10s
CF (IL IV,VI) 4 0 (± 12)xl02 7 5 (±2 8)xl03' 9 0 (± 3 1) x 105 1 9 x 104 71 xlO4 4 1 (±0 3)xlOs
EF (II, IV, VI) 3 3 (± 13)xl02 4 6 (± 1 2) x 104' 7 8 (± 1 9) x 105 1 5 x 105 41 xlO5 4 5 (±0 4)xlOs
TMP (TV, V) 3 3 (±3 0)xl04 7 4 (±1 8)xl04 5 2 (± 1 0) x 105 2 7 x 105 43 xlO5 6 9 (±0 2)xlOs
LM(V) NA7 3 3(±0 1)xl05(23) 2 8 (± 0 1) x 106 (23) 6 7 x 105 14 xlO6 8 5 (±0 2)xlOs
PG (II, V) NA^ 4 8(±0 1)xl03j 4 8(±0 1)xl03j 4 8 x 103 48 xlO3 7 3 (±0 3) x 10s"1
cp (ni, v) nl/ 8 2 (±2 9)xl04 9 3 (± 2 2) x 104 8 7 x 104 91 xlO4 8 5 (±0 7)xlOs
TET (in, VI) 9 4 (±0 6)xl04 -47(±03)xl06(pH 3 to 9, see Figure if 1 9 x 106 32 xlO6 7 7 (± 12)xlOs
VM (TV, V) ll(±01)xl04 -9 1(±ll)xl05(pH 3 to 8, see Figure if 6 1 x 105 81 xlO5 8 1 (±0 3)xlOs
AM (L VII) 13 (±0 7)x 101 - 1 1 (± 0 2) x 104 (pH 4 to 9) see Figure if 1 8 x 103 49 xlO3 7 2 (±0 3)xlOs
^Values obtained from the current investigation, unless indicated otherwise. ^For
full names, see Table 1. "Described in Table 3. For CF, II was used from pH 3-6,and IV from pH 6.5-8. For EF, II was used from pH 3-5.5, and IV from pH 6-
8. ^20(±0.5) °C. '25(±0.5) °C. JNA - Not applicable, ND - Not determined.
•sProtonated amine reactivity assumed to be negligible, on the basis of prior
observations (21,22). *Only ranges of k o3,apP are listed for these compounds. *
k o3 for the "monoprotonated" fluoroquinolone species represents an
"effective" rate constant for the combination of zwittenonic and neutral species.
-'PG reactivity assumed to be independent of acid-base speciation, because its
dissociable functional groups are not conjugated with its thioether (Table 1).^Rate constant recalculated with pKa,RX = 9.2, using data reported by Huber et
al. (17). The small difference between AZ's pKai and pKa2 — combined with lack
of data at pH > 6.8 — prevented accurate determination of k o3,apP for neutral
AZ. This value agrees well with that reported by Phillips et al. (29) - corrected
according to Neta and Dorfman (1968) (30) (k oH,apP,PG= 7.1 x 109 Mh 1).
Moiety-specific oxidation of antibacterial molecules 171
106 -
105 -
104 -
103 -
102 -
101 H
(a)
O app TYL
*y ^
^^
/o o
k //0 app AZ /7
//
'V
O O app TYL
D 0 app AZ
10
PH
105 -
JT^ 104 -
d 103 ^
j/ 102
101
10°
O *0 app SMX V O app EF
O O app ASMX D "o app CF
PH
10
172 Appendix I
(c)k
a
"•0 app LMD *0 app TMP
O 0 app CP
" *0 app PG106 -
"•0 app TMT
105 -
^Oyf-~\
\ *oapp CP
*'k." "'O app MP
104 - *oapp TMP
"•0 app PG
m3 -
•** k
"•0 app DAMP
10
PH
Figure 1. Apparent second-order rate constants for reactions of
parent substrates and associated substructure model substrates with
O3 at 20(_t0 5) °C (symbols = measurements, lines = model
predictions) (a) macrolides (RX, AZ, and TYL) with associated
substructure models, (b) sulfonamides (SMX and ASMX) and
fluoroquinolones (CF and EF) with associated substructure models,
(c) TMP, LM, and ß-lactams (PG and CP) with associated
substructure models ^k 03 app RXand k 03 app LM
calculated from data
reported by Huber et al (17) and Qiang et al (23), respectively
14 I 2 Sulfonamides
Sulfamethoxazole (SMX). The apparent rate constants measured for
reaction of SMX with O3 range from ~5 x 104 to ~5 x 105 M h 1
between pH 3 and 7 (Figure lb) The specific rate constant calculated
for the SMX anion (Table 4) agrees withm a factor of 2 2 with that
previously determined by Huber et al (17), after correction for the
different O3 consumption stoichiometnes of the reference competitor
substrates — cmnamic acid (28) and phenol (24,33) — used in the current
and former studies Neutral 4-ammophenyl methyl sulfone (APMS in
Table 2) — which approximates SMX's aniline moiety — reacts with 03
with the same rate constant calculated for the neutral SMX species
(Table 4, Figure lb) In contrast, 3,5-dimethyhsoxazole (DMI in
Moiety-specific oxidation of antibacterial molecules 1 73
Table 2) — a model for SMX's isoxazole group — reacts very slowly with
O3 (Figure lb). These results indicate that reaction of O3 with the SMX
structure takes place primarily at the biochemically-active p-sulfonyl
aniline moiety. The pH-dependency of k"o3japBsMx (Figure lb) appears
to correlate with deprotonation of SMX's sulfonamide-mtrogen
(Table 1). However, rapid reaction between O3 and the sulfonamide-
mtrogen can be ruled out (21,22). Therefore, this effect seems to be
due to activation of the SMX molecule's aniline moiety (Table 1)
toward electrophilic attack by O3, via deprotonation of the
sulfonamide-mtrogen.
7V(4)-acetyl-sulfamethoxazole (ASMX). A high percentage of SMX
enters wastewater treatment facilities as the metabolite ASMX (Table 1)
— which can be retransformed to SMX during biological treatment (4).
The rate constants determined for reaction of ozone with ASMX's
neutral and anionic species are 2.0 x 101 and 2.6 x 102 M h1,
respectively; more than three orders of magnitude lower than for SMX
(Figure lb). These results agree with prior observations that ASMX is
poorly degraded during ozonation of secondary wastewater effluent (18).
The decrease in reactivity from anionic ASMX to neutral ASMX (Table
4, Figure lb) likely reflects enhancement of the sulfonamide moiety's
electron-withdrawing strength upon sulfonamide-mtrogen protonation.
Because this effect should reduce the 03-reactivities of both the ASMX
isoxazole ring and the p-svlionjl aniline ring, it is unclear which moiety
dominates observed reaction kinetics (Figure lb).
1.4.1.3 Fluoroquinolones.
Ciprofloxacin (CF). CF's reactivity toward O3 is strongly dependent
on pH, where k"o3japBcF ranges from ~2 x 102 to more than 105 M h 1
between pH 3 and 8 (Figure lb). The pH-dependency of k"o3,aPBcF
indicates that this trend is governed by deprotonation of CF's N(4)
amine (Table 1). This hypothesis is supported by the reactivity of O3
174 Appendix I
with ethyl N-piperazmecarboxylate (EPC in Table 2). k"o3,aPBEPC
exhibits nearly the same pH-dependency and magnitudes as k"o3,aPBcF
between pH 5 and 8 (Figure lb). Flumequme (FLU m Table 2) — a
surrogate for the biochemically-active qumolone moiety (Table 1) —
also exhibits pH-dependent reaction kmetics. However, values of
k"o3,a.pp,FLu are several orders of magnitude lower than those observed
for CF's N(4) amine (Figure lb). The baseline reactivity observed for
the CF molecule at pH < 4 (Figure lb) cannot be attributed to
reactions with the N(4) atom or the qumolone heterocyclic rmg
(Figure lb). This reactivity must therefore be due either to reactions
taking place at the piperazme N(l) atom or at the unsubstituted
ortho-position of the ad|acent aromatic rmg
Enrofloxacin (EF). EF reacts with O3 at higher rates than CF m the pH
région between 3 and 8 (Figure lb). This effect is due pnmanly to the
difference m CF's and EF's pKa2 values (Table 1). At pH 7, for example,
the molar fraction of anionic CF (m which the N(4) amme is
deprotonated) is 0.01, compared to 0.15 for EF. However, the apparent
rate constants for CF and EF converge at higher pH (Figure lb),
indicating a close similanty m the absolute reactivities of their N(4) atoms.
1.4.1.4 Trimethoprim (TMP].
Measured magnitudes of k"o3,aPBTMP range from high-104 to
mid-105 M h1 (Figure lc). The observed vanation m k"03japBTMP
(Figure lc) can most likely be attnbuted to speciation of its
diammopynmidme moiety (Table 1). Protonation at the heterocyclic
N(l) and N(3) nitrogens (Table 1) should reduce this moiety's O3
reactivity via coordination of the resonant lone-pair electrons
associated with each of TMP's two exocyclic amines. 2,4-diammo-5-
methylpynmidme (DAMP m Table 2) also exhibits pH-dependent
vanation m its apparent O3 reaction rate constant (Figure lc). However,
the specific rate constants calculated for mono- and diprotonated
Moiety-specific oxidation of antibacterial molecules 1 75
DAMP (Table S3) are substantially lower than those calculated for the
corresponding TMP species (Table 4, Figure lc). The high O3 reactivity
of 3,4,5-tnmethoxytoluene (TMT m Table 2) — a surrogate for TMP's
tnmethoxytolyl moiety (Table 1) — suggests that the tnmethoxytolyl
moiety accounts for the high reactivity of the protonated TMP species
(Figure lc). The difference m reactivities of TMT and TMP's protonated
species could be a consequence of the TMP diammopynmidme moiety's
bulk, which may hmder attack by O3 at the 2- and 6-positions of the
TMP structure's tnmethoxytolyl moiety (Table 1).
1.4.1.5 Lincomycin (LM).
k"o3,aPBLM is constant below pH 5, and mcreases above pH 5 to a
calculated maximum of 2.8 x 106 M h 1 (Figure lc). LM's baselme
reactivity can be attnbuted to pH-mdependent kmetics of the reaction
between its thioether and O3, smce its heterocyclic amine is protonated— and essentially unreactive toward O3 — under these conditions (23).
Likewise, the vanation in k"o3,aPBLM above pH 5 can be attnbuted to
reaction of O3 with the neutral heterocyclic amme (23). k"03japp values
measured for MP — a model for LM's heterocyclic tertiary amine
(Table 1) — are consistent with these conclusions (Figure lc).
1.4.1.6 ß-lactams.
Penicillin G (PG). PG possesses only one functional moiety - a
thioether (Table 1) - that can be expected to account for its observed
reactivity toward O3 (Figure lc). Reaction kinetics for this moiety are
expected to be independent of pH (22,23).
Cephalexin (CP). k"03jilpBcp is a little more than one order of
magnitude higher than k"o3,aPBPG on average (Figure lc). The relatively
high magnitude of k"o3,aPBcp at pH < 7 suggests that reactivity of CP at
acidic pH is attnbutable either to oxidation of its double bond or
thioether (Table 1), each of which is expected exhibit pH-mdependent
176 Appendix I
O3 reaction kmetics. However, the slight increase m k"o3,aPBcp above
pH 7 (Figure lc) suggests that the primary amine (pK* = 7.1, Table 1)
may govern CP reactivity at circumneutral and higher pH values.
1.4.1.7 Substrates with more than two pKa values.
The complexity of TET's, VM's, and AM's speciation patterns
precluded accurate modeling of O3 reaction kmetics by the approaches
utilized above. However, k"o3,<iPBTET, k"o3,<iPBvM, and k"o3,<iPBAM were
compared to substructure model substrate data (Figure 2) to facilitate
preliminary assignment of moiety-specific O3 reaction kmetics.
"•O, app TMT_
TT
0 0
10" ! £&"/*105
!
104!
=0=^^ 03 app MBDCH
ks*"'''*** "'
"•Oj app DMCH J^ . p.*
<*> .••"' >..••'!•'' \c ..' %oy'y- i0< m CH
103!
102-|
D
O
O
k"•Oj app TET
le"•03 app VM
k"•Oj app AM
101-| ?
> 4 6
PH
8 1
Figure 2. Apparent second-order rate constants for reactions of TET,
VM, AM, and associated substructure models with O3 at 20(+0 5)°C
(symbols = measurements, lines = model predictions)
TET reacts rapidly with O3 over a wide range of pH (Figure 2).
Because the rate of reaction with the tertiary amine (pKa = 7.7,
Table 1) is not likely to be appreciable relative to the remainder of the
TET molecule below pH 5 (Figure 2), the TET structure's observed
reactivity toward O3 at acidic pH is likely due to oxidation of the
tetracyclme nng system. Although tertiary amme reactivity could be
Moiety-specific oxidation of antibacterial molecules 177
important above pH 7, the high reactivity of vanous phenolic
structures (21,24) and 2-(3-methylbutyryl)-5,5-dimethyl-l,3-
cyclohexandione (MBDCH m Table 2) — a surrogate for the olefimc
bonds withm TET's tetracyclic rmg system (Table 1) — indicate that
O3 reacts pnmanly with the rmg system at circumneutral pH, as well.
VM is also highly reactive toward O3 between pH 3 and 8 (Figure 2).
The high measured reactivity of TMT (Table 2, Figure 2) — a
surrogate for VM's tnmethoxybenzyl moiety, in addition to the high
known reactivity of phenols and resorcmols toward O3 at all pH
conditions (21,24), suggests that k"o3,aPBvM at pH < 7 corresponds to
oxidation of the biochemically-essential aromatic target sites shown
in Table 1, since oxidation of VM's amine moieties should not be
important at acidic pH. However, the secondary N-methylleucme
moiety may react rapidly enough with O3 to influence observed
reaction kinetics at pH > 7, on the basis of known secondary amine
reaction kmetics (21,22).
k"o3,aPBAM vanes from roughly 101 to |ust over 104 M h l between
pH 2 and 8.6 (Figure 2). The measured pH-dependency of k"o3,aPBAM
is consistent with oxidation of AM's primary amine moieties (with
pKa values ranging from 6.7 to 9.7, Table 1). In addition, k"o3,aPBAM
near pH 9 (at which all but one of AM's amines are predominantly
deprotonated) is of the same order of magnitude as k'03 for neutral
butylamme (1.2 x 105 (22)), cyclohexanemethylamme (CHM in Table
2, with 7.1 x 104, from Table S3), and cyclohexylamme (CH in Table
2, with 4.9 x 104 M h 1, from Table S3) — which can be taken as
structural approximations of AM's primary amine groups (Table 1).
178 Appendix I
1.4.1.8 Moiety-specific Oxidation vs. Parent Molecule
Disappearance
The preceding discussion indicates that initial reactions of O3 with
many parent antibactenal substrates occur at the substrates' essential
target moieties (Table 1). Consequently, ti/2 values for observed
transformations of many parent molecules correspond to ti/2 for their
essential target moieties (Figure Sil). However, ti/2 for reaction of O3
with FLU — which approximates CF's and EF's biochemically-essential
qumolone moieties — is 14 times longer at pH 7 than t,i/2,cf (which
corresponds to oxidation of the nonessential piperazme moiety) and
109 times longer than ti/^EF (Figure Sil), indicating that observed
losses of parent fluoroquinolones dunng ozonation may not necessanly
correspond directly to elimination of their antibactenal activities. In
addition, O3 does not appear to oxidize essential targets m the ASMX,
PG, or CP molecules at appreciable rates. However, the
03-recalcitrance of these three compounds' essential target moieties
may be of relatively minor importance, since pnor findmgs mdicate
that ASMX and ß-lactams m general are unlikely to be discharged to
surface waters at significant concentrations (1,4).
I.4.2 Studies in Wastewater Matrixes
I.4.2.1 Depletion of Parent Substrates
Oxidation of 1 |oM ASMX — the substrate reacting slowest with O3
(Table 4) — was monitored m batch expenments with Kloten-Opfikon
wastewater at an O3 dose of 63 |oM (3 mg/L) (Figure S12).
Approximately 35% of [ASMXJo remained after nearly complete
depletion of O3. The observed recalcitrance of ASMX is consistent
with observations for pilot-scale ozonation of Kloten-Opfikon
secondary effluent (18).
Moiety-specific oxidation of antibacterial molecules 179
Measured losses of fast-reacting (i.e., k"o3,aPBM > 103 M asa)
antibactenal substrates dunng ozonation of Kloten-Opfikon
wastewater were > 99% at O3 doses > 3 mg/L (63 |xM) (Figure 3a).
TET was > 99% transformed at an O3 dose of 1.5 mg/L (31 |xM)
(Figure 3a), consistent with the high magnitude of k"o3,aPBTET at pH 7.7
(Table 4). However, significant residuals of PG remained even at an O3
dose of 3 mg/L (63 |xM) (Figure 3a), consistent with the low
magnitude of k"o3,aPBPG. Nmety-nme % PG loss was only achieved at
an O3 dose of 5 mg/L (104 |xM). Observed losses of SMX (Figure 3a)
agree well with results reported for pilot-scale ozonation of Kloten-
Opfikon wastewater at an O3 dose of 1 mg/L (18). RX, AZ, AM, and
LM were not mcluded m these expenments because of analytical
difficulties. Predicted losses for these four substrates were calculated
for an applied O3 dose of 1 mg/L (Figure 3a), as descnbed m Text S5.
The values calculated for RX and AZ (Figure 3a) agree very well with
pilot-scale measurements under similar conditions (18).
100
£ $ #£&*£
Applied 03 Dose
VWA 0 5 mg/LI 1 1 mg/LFS^sa 1 5 mg/LI I 3 mg/L^^M 5 mg/L
K^ ^
180 Appendix I
100
80
c
o
1 60
,o</>c
ra 40
H
20 \
(b)* '
100
Q 80
E 60
bw 40
05
0
TET. TYL
«* f"»" 4m*o •
a VMu"'
.RX*P
.
CF
AM^CP
PG
TET
VMTYL 0
SMX \ \
^\„„,
tV LM*
1 4 6 8 1
*.0H=pp"><1°9(M1s,)
AM*
PG
\ AZ*0 ^cp TMP
CF
# Measured
O Estimated
101 102 103 104 105 107
Figure 3. Antibacterial substrate transformation during ozonation of
Kloten-Opfikon wastewater at 20(±0 5) °C, pH 7 7, and
[substrate]o = 1 uM (a) Measured transformation efficiencies at
varying O3 dosages, for substrates with k 03 app> 103 M 1s 1
(b) Dependence of transformation efficiencies on k 03 app,at an O3
dose of 1 mg/L (21 uM) Inset shows independence of transformation
efficiency from k .oHapp *% transformation for RX, AZ, LM, and AM
estimated via 1 edures described in Text S5
1.4.2.2 Role of O3 and 'OH in Oxidation of Parent Substrates
Substrate transformation dunng a real water ozonation process can be
charactenzed by eq 3 (34),
InM[Ml 03,app,M
\0jj[03]^-£.OHiapPiMj[.OH>/f (3)
where the two integral terms in eq 3 represent the O3 and *OH
exposures (26,34) governing transformation of substrate, M, over
reaction time, t, where [*OH] = fit) and [O3] = ^(t). k".oH,aPBM values
measured for each antibactenal compound range from 2.9 to
8.5 x 109 M h 1 at pH 7 (Table 4).
Moiety-specific oxidation of antibacterial molecules 181
As illustrated m Figure 3b, the transformation efficiencies determined
for each substrate m Kloten-Opfikon wastewater correlate well with
magnitude of k"o3,aPBM- In contrast, there is no clear relation between
transformation efficiency and k".0H,aPBM for these conditions (mset to
Figure 3b). This indicates that [M]/[M]o is governed pnmanly by the O3
oxidation term m eq 3. The »OH oxidation term becomes important
only if the magnitude of the O3 oxidation term is relatively small,
either as a consequence of low k"o3,aPBM, low JOsdt, or high J'OHdt.
Contnbutions of O3 and «OH to oxidation of ASMX were evaluated
according to eq 4,
( "
(4)OH app Mk.^.,...Ml[-OH]dt
f
1
\^
OH]#/j[0/ 0 J)
k0, Spp„|[03^+fcoHSpp„|[-OH]j/
where £oh,m(t) represents the fraction of substrate oxidation due to
•OH after reaction time, x. JOsdt was obtained by direct measurement
of O3 (via the mdigo method (35)), and J'OHdt was determmed from
measured loss of the »OH probe pCBA, which reacts rapidly with »OH
and very slowly with O3, accordmg to eq 5 (34).
In
y[pCBA]OJ-^oh^cbaJl-OH]^ (5)
These calculations mdicated that ASMX was transformed exclusively by
•OH (Figure S12).
Transformations of fast-reacting substrates were also charactenzed
according to the relationships shown in eq 4. However, because JOsdt
could not be measured dunng the course of each substrate's
182 Appendix I
transformation m Kloten-Opfikon wastewater, f.oH,M(x) was actually
calculated from eq 6 (18), which is obtamed by substitution of eqs 3
and 5 into eq 4.
•OH,app,MIn
f.•OH,app,pCBA
[^CBA]r[pCBA]
to J
OH,M f
In[MLMo
(6)
Although values obtamed via eq 6 provide no mformation as to temporal
vanation of JOsdt dunng ozonation, this expression can still provide
reliable estimates of the absolute contributions of «OH and O3 to bulk
substrate oxidation. For example, £ohjm(x) values calculated by eq 6 for
the antiepileptic drug carbamazepme (with k"o3,app ~ 3 x 105 M h 1,
k'.oH = 8.8 (± 1.2) x 109 M h 1 at pH 7 (17)) typically exhibit deviations
of less than 10% from values determmed by eq 4, usmg measurements
of O3 and «OH exposure actually taken withm the first few hundred
milliseconds after application of O3 to vanous municipal wastewaters
(Figure S14). £oh,m(t) values calculated by eq 6 are summanzed m Figure
4a. According to eq 4 (from which eq 6 is denved), £oh,m(t) should be
mversely related to the ratio k'c^aj^M/k'.oHicßM, consistent with the
general trend shown m Figure 4a.
Moiety-specific oxidation of antibacterial molecules 183
Figure 4. Importance of O3 and -OH in transformation of fast-reactingsubstrates (ko3app > 103 M1s1) dunng ozonation of Kloten-Opfikon
secondary effluent at 20(±0 5) °C, pH 7 7, [substrate]0 = 1 |iM, and
[O3]o = 21 |J,M (1 mg/L) (a) Calculated fractions of observed substrate
transformation due to O3 and •OH (b) Correlation of measured values of
f>OHM(x) (obtained via eq 6) with values calculated via eq 4 for the range of
Rctfv) values (25) expected for Kloten-Opfikon wastewater at these conditions
T corresponds to reaction tune after O3 dosage R^ represents the cuniulative
ratio of -OH exposure to O3 exposure for a pollutant molecule after reaction
ùme, T *Esùmates of % transformation and f-oHM(x) f°r R^ AZ> LM, and
AM were calculated via procedures descnbed in Text S5
184 Appendix I
1.4.2.3 Effects of Matrix Characteristics on Relative Importance
of O3 and 'OH in Observed Substrate Transformation
As evident from eq 4, £oh,m(x) for a given substrate will also depend on
the time-dependent ratio of /•OHdt/Jb3dt, or R^ (25,33,34,36)
dunng wastewater ozonation. Rct(x) is particularly sensitive to
wastewater DOC concentrations, where higher DOC typically
translates to higher R^) values (25,36). The solid lmes shown in Figure
4b illustrate expected £oh,m(x) values (calculated from eq 4) as a
function of k"o3japBM/k".oH,aPBM, for a number of R^x) values. The
shaded region shown m Figure 4b represents the apparent range of
R<:t(x) values previously observed (at O3 doses varying from 1 to
2.5 mg/L) for vanous municipal wastewaters dunng the time-scales
(< 10 s) withm which the bulk of fast-reacting substrate loss
(i.e., > 90%) is expected to occur. In accordance with these
expectations, the £oh,m(x) values obtained by eq 6 (for an applied O3
dose of 1 mg/L) for each fast-reacting substrate mcluded m the present
investigation fall withm this shaded region (Figure 4b). The
relationships presented m Figure 4b suggest that £oh,m(x) for substrates
with values of k^vppjvi/k'.oH^j^M < 10 6 will range from about 0.5 to 1
dunng ozonation of a typical wastewater. However, £oh,m(x) can be
expected to fall anywhere between 0.1 and 0.9 for substrates with
k"o3,a.pBM/k".oH,a.Pp>i ranging from 10 6 to 10 4. In contrast, £oh,m(x) will
generally be relatively small (< 0.5) for substrates with
k 03,aPBM/k •OH,aPBM> 10 4.
1.4.2.4 Implications for Selective Oxidation during Wastewater
Ozonation
The ma|onty of antibactenal substrates mcluded m this study are
expected to be transformed predominantly via direct oxidation by O3
dunng wastewater ozonation (Figure 4b). However, PG, CP, AM, and
Moiety-specific oxidation of antibacterial molecules 185
ASMX (each with k"o3,aPP,M/k".oH,aPBM < 104) will generally be
transformed to a large extent by »OH dunng wastewater ozonation.
This may be undesirable in the case of AM, smce *OH will likely react
mdiscnmmately with many sites not associated with the AM molecule's
antibactenal activity (Table 1 and Text SI). With respect to ASMX,
however, »OH may be no less desirable as an oxidant than O3, because
O3 appears to react only very slowly (if at all) with the essential
p-svlfonyl aniline target (Table 1, Table 4). Oxidation of PG and CP by
•OH may actually be more desirable than oxidation by O3, since »OH
reactions lead pnmanly to production of biochemically-inactive (37)
benzylpemlloic and benzylpemcilloic acids (29). In the case of CF and
EF, the relatively low ratios of kœ^j^target (i.e., k"03,<iPBFLu) to k'.oH^^M
(i.e., < 106, according to Figure lb and Table 4) suggest that
nonessential locations withm each fluoroquinolone's structure may be
oxidized to a significant extent by «OH and O3 pnor to reaction of O3
with their essential target moieties. On the assumption that pre-
oxidation of other locations withm the parent structures (by O3 or
•OH) does not significantly reduce the reactivity of their target
moieties toward O3, full oxidation of the target moieties can likely still
be achieved by selecting applied O3 dose to mamtam a sufficiently high
JOsdt, which should be chosen on the basis of moiety-specific
oxidation kinetics, as opposed to observed parent substrate
transformation kinetics.
1.5 Acknowledgements
Michael Dodd gratefully acknowledges financial support from a U.S.
National Science Foundation Graduate Research Fellowship. The
authors thank Marc Huber, Gretchen Onstad, Andreas Peter, Silvio
Canomca, and Yunho Lee for helpful discussions, and Elisabeth Salhi
for technical assistance. Two anonymous reviewers are thankfully
acknowledged for their constructive comments.
186 Appendix I
1.6 Supporting Information Available
Text, Figures, and Tables addressing matenals, expenmental
procedures, substrate reactive sites, and biochemical mechanisms of
antibactenal activity This matenal is available free of charge via the
Internet at http://pubs.acs.org
1.7 Literature Cited
1 Hirsch, R, Ternes, T A, Haberer, K, Kratz, K -L, Occurrence of
antibiotics in the environment Set. Total Environ. 1999, 225, 109-118
2 Golet, E M , Xifra, I, Siegnst, H , Alder, A C, Giger, W, Environmental
exposure assessment of fluoroquinolone antibacterial agents from sewage
to soil Environ. Sa. Technol. 2003, 37, 3243-3249
3 Miao, X-S, Bishay, F, Chen, M, Metcalfe, C D, Occurrence of
antimicrobials in the final effluents of wastewater treatment plants in
Canada Environ. Sa. Technol. 2004, 38, 3542-3550
4 Gobel, A , Thomsen, A , McArdell, C S, Joss, A , Giger, W, Occurrence
and sorption behavior of sulfonamides, macrohdes, and trimethoprim in
activated sludge treatment Environ. Sa. Technol. 2005, 39, 3981-3989
5 Lindberg, R H, Wennberg, P, Johansson, M I, Tysklind, M , Andersson,
B A V, Screening of human antibiotic substances and determination of
weekly mass flows in five sewage treatment plants in Sweden Environ. Sa.
Technol. 2005, 39, 3421-3429
6 Kim, S, Eichorn, P, Jensen, J N , Weber, A S
, Aga, D S,Removal of
antibiotics in wastewater effect of hydraulic and solid retention times on
the fate of tetracycline in the activated sludge process Environ. Sa. Technol.
2005, 39, 5816-5823
7 Brain, R A, Johnson, D J, Richards, S M
, Hanson, M L, Sanderson,
H, Lam, M W, Young, C, Mabury, S A, Sibley, P K, Solomon, K R,
Microcosm evaluation of the effects of an eight pharmaceutical mixture
to the aquatic macrophytes Eemna gibba and Mjriophyllum sibinaim Aquatic
Toxicology 2004, 70, 23-40
8 Wilson, C J, Structural and functional responses of plankton to a
mixture of four tetracyclines in aquatic microcosms Environ. Sei. Technol.
2004, 38, 6430-6439
9 Wilson, B A, Smith, V H , Denoyelles, F, Lanve, C K
,Effects of three
pharmaceutical and personal care products on natural freshwater algalassemblages Environ. Sa. Technol. 2003, 37, 1713-1719
Moiety-specific oxidation of antibacterial molecules 187
10 Lonan, V, Ed Antibiotics in Eaboratory Mediane, 4th ed, Williams and
Wilkins Baltimore, MD, 1996
11 Baquero, F, Low-level antibacterial resistance a gateway to clinical
resistance Drug Resistance Updates 2001, 4, 93-105
12 Drkca, K ,The mutant selection window and antimicrobial resistance J.
Antimicrob. Chemother. 2003, 52, 11-17
13 Paraskeva, P, Graham, N J D, Ozonation of municipal wastewater
effluents Water Environ. Res. 2002, 74, 569-581
14 Beltran, F J, Garcia-Araya, J F, Alvarez, P M, Integration of continuous
biological and chemical (ozone) treatment of domestic wastewater 2
Ozonation followed by biological oxidation /. Chem. Technol. Biotechnol.
1999, 74, 884-890
15 Gehr, R, Wagner, M, Veerasubramanian, P, Payment, P, Disinfection
efficiency of peracetic acid, UV and ozone after enhanced primary
treatment of municipal wastewater Water Res. 2003, 37, 4573-4586
16 Adams, C, Wang, Y, Loftin, K, Meyer, M ,Removal of antibiotics from
surface and distilled water in conventional water treatment processes J.Environ. Eng. 2002, 128, 253-260
17 Huber, M M, Canonica, S, Park, G-Y, von Gunten, U, Oxidation of
pharmaceuticals during ozonation and advanced oxidation processes
Environ. Sa. Technol. 2003, 37, 1016-1024
18 Huber, M M, Gobel, A , Joss, A, Hermann, N, Loffler, D, McArdell, C
S, Ried, A, Siegrist, H, Ternes, T A, von Gunten, U, Oxidation of
pharmaceuticals during ozonation of municipal wastewater effluents a
pilot study Environ. Sei. Technol. 2005, 39, 4290-4299
19 Ternes, T A, Stuber, J, Herrmann, N, McDowell, D, Ried, A,
Kampmann, M, Teiser, B, Ozonation a tool for removal of
pharmaceuticals, contrast media and musk fragrances from wastewater^
Water Res. 2003, 37, 1976-1982
20 Huber, M M, Ternes, T, von Gunten, U, Removal of estrogenic activity
and formation of oxidation products during ozonation of 17a-
ethinylestradiol Environ. Sa. Technol. 2004, 38, 5177-5186
21 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water — II Dissociating organic
compounds Water Res. 1983, 17, 185-194
22 Pryor, W A, Giamalva, D H, Church, D F, Kinetics of ozonation 2
Amino acids and model compounds in water and comparisons to rates in
nonpolar solvents J. Am. Chem. Soc. 1984, 106, 7094-7100
23 Qi^ng, 2 , Adams, C, SurampaUi, R, Determination of ozonation rate constants
for lincomyan and specunomycin O^one: Sa. Eng. 2004, 26, 525-537
188 Appendix I
24 Mvula, E, von Sonntag, C, Ozonolysis of phenols in aqueous solution
Org. Biomo/. Chem. 2003, /, 1749-1756
25 Buffle, M -O, Schumacher, J, von Gunten, U, Ozonation and advanced
oxidation of wastewater Effect of O3 dose, pH, DOC and HOt-
scavengers on ozone decomposition and HO" generation O^one: Sa. Eng.,
accepted, 2006.
26 von Gunten, U, Ozonation of drinking water Part I Oxidation kinetics
and product formation Water Res. 2003, 37, 1443-1467
27 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water — I Non-dissociating organic
compounds Water Res. 1983, 17, 173-183
28 Leitzke, A , Reisz, E , Flyunt, R, von Sonntag, C ,The reactions of ozone
with cinnamic acids formation and decay of 2-hydroperoxy-2-
hydroxyacetic acid /. Chem. Soc, Perkm Trans. 2 2001, 793-797
29 Phillips, G O, Power, D M, Robinson, C, Chemical changes following
y-irradiation of benzylpenicillin in aqueous solution J. Chem. Soc, Perkm
Trans. 21973,575-582
30 Neta, P, Dorfman, L M,Pulse radiolysis studies XIII Rate constants
for the reaction of hydroxyl radicals with aromatic compounds in
aqueous solutions Adv. in Chem. 1968, 81, 222-230
31 Dowideit, P, von Sonntag, C, Reaction of ozone with ethene and its
methyl- and chlorine-substituted derivatives in aqueous solution Environ.
Set. Technol. 1998, 32, 1112-1119
32 Onstad, G D, Strauch, S, Meriluoto, J, Codd, G, von Gunten, U,
Selective oxidation of key functional groups in cyanotoxins during
drinking water ozonation Environ. Sa. Technol., submitted, 2005
33 Buffle, M -O, Schumacher, J , Salhi, E ,
von Gunten, U, Measurement of
the initial phase of ozone decomposition in water and wastewater bymeans of a continuous quench flow system Application to disinfection
and pharmaceutical oxidation Water Res., accepted, 2006.
34 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios duringozonation processes I The Rct concept O^one: Sa. Eng. 1999, 21, 239-260
35 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Water Res. 1981, 15, 449-456
36 Elovitz, M S, von Gunten, U, Kaiser, H P, Hydroxyl radical/ozone
ratios during ozonation processes II The effect of temperature, pH,
alkalinity, and DOM properties O^one: Sa. Eng. 2000, 22, 123-150
37 Walsh, C Antibiotics:Actions, Origins, Resistance, ASM Press Washington, DC, 2003
All Oxidation of Antibacterial Molecules by
Aqueous Ozone: Moiety-Specific Reaction
Kinetics and Application to Ozone-Based
Wastewater Treatment
Michael C Dodd, Marc-Ohvier Buffle, Urs von Gunten
Environmental Saence and Technology, 2006
Supporting Information
Texts
51. Mechanisms of antibacterial compounds' biochemical activities and relevance to
target sites at which O3 is expected to react with parent antibacterialmolecules.
52. Chemical reagents.
53. Measurement of apparent second-order rate constants for O3 and -OH reactions
54. Wastewater matrix experiments.
55. Estimation of transformation efficiencies and foHM for RX, AZ, AM, andLM
Tables
51. pKa values and corresponding source references for each antibacterial substrate
52. pi*Q values and corresponding source references for each substructure model substrate
53. Second-order rate constants (MV1) for reactions of O3with substructuremodel substrates
Figures
51. Maximum single-compound concentrations of vanous antibactenal classes detected
in municipal wastewater systems and surface waters, in the context of minimum
reported clinical MIC values for sensitive bacterial reference strains.
52. Biochemicalmodel for mechanism of macrolide antibacterial activity
53. Biochemicalmechanism of sulfonamide antibactenal activity
54. Biochemicalmodel for mechanism of fluoroquinolone antibacterial activity
55. Biochemicalmechanism of dihydrofolate reductase (DHER) inhibitor antibactenal activity
56. Biochemicalmodel for mechanism of lincosamide antibacterial activity
57. Biochemicalmechanism of /^lactam antibactenal activity (depicted for penicillin G)58. Biochemicalmodel for primarymechanism of tetracycline antibacterial activity
59. Biochemicalmodel for mechanism of glycopeptide antibacterial activity
S10. Biochemical model for mechanism of aminoglycoside antibacterial activity
Sil. Calculated ti/2 values for the apparent transformation of model antibacterial substrates
by O3, in companson to corresponding estimated half-lives for reaction of O3 with the
targeted functional moieties associatedwith each substrate's biochemical activity
512. Transformation of ASMX dunng ozonation of Kloten-Opfikon wastewater
513. Correlations of predicted substrate transformation with measured values
S14 Companson of f.oHM(x) values calculated from indirect determinations of O3
exposure with those calculated from direct measurements of O3 exposure
190 Appendix II
11.1 Texts
11.1.1 SI. Mechanisms of antibacterial compounds' biochemical
activities and relevance to target sites at which O3 is
expected to react with parent antibacterial molecules.
II. 1.1.1 Macrolides
Macrohde antibactenals are believed to denve their biochemical activity
from specific hydrogen bonding with vanous nucleobases and
phosphodiester linkages in the peptidyl transferase cavity of bactenal 23S
rRNA (Figure S2) (1). The bonding interaction most Hkely to be
interrupted by direct reaction with O3 is that involving each macrolide's
charactenstic tertiary amine (Figure S2), which is expected to be highly
reactive toward O3 (2,3). Oxidation of the tertiary amine via formation of
an aminoxide or via demethylation (3) should prevent its hydrogen bonding
with 23S rRNA, leading to reduction or elimination of each parent
macrolide's antibactenal activity Macrolides with fifteen- and sixteen-
membered macrolactone rings generally possess additional moitiés that can
be expected to react rapidly with O3. For example azithromycin (AZ — a
fifteen-membered macrolide — shown in Table 1 within the main text) and
tylosin (TYL — a sixteen-membered macrolide — shown in Table 1) contain
an additional tertiary amine and a con|ugated diene, respectively. Oxidation
of azithromycin's (AZ) heterocyclic nitrogen would likely be sufficient to
impair the parent structure's antibactenal activity if such a reaction resulted
in rupture of its macrolactone ring (e.g., via dealkylation of the nitrogen
atom (3J). This would presumably result in interruption of the specific
stereochemistry necessary for appropnate hydrogen bonding between the
AZ structure and bactenal rRNA (Figure S2). Similarly, reaction of O3 with
tylosin's (TYL) diene moiety should lead to impairment of the parent
structure's antibactenal activity, where ozonolysis of one or both olefinic
bonds would result in rupture of TYL's macrolactone ring (4).
Oxidation of antibactenals: supporting information 191
11.1.1.2 Sulfonamides
All sulfonamide antibactenal structures are denvatives of
^-aminobenzenesulfonamide — a structural analog of ^-aminobenzoic
acid (pABA) — and are differentiated only by the particular
R-substituent attached to the sulfonamide nitrogen (Figure S3). These
compounds denve their antibactenal activity from antagonistic
competition with pABA for diliydropteroate synthase enzyme dunng
bactenal synthesis of dihydropteroic acid (the precursor to folic acid)
(Figure S3) (5). The /»-sulfonyl aniline moiety, in particular, is
responsible for each sulfonamide's interference with bactenal folate
synthesis (Figure S3). This moiety should present a very favorable
target for O3, since aromatic amines are generally very reactive toward
O3 (2,6,7). Sulfonamides' R-substituent moieties (e.g., isoxazole in the
case of sulfamethoxazole, SMX — shown in Table 1 withm the main
text), can also be expected to react with O3. These latter reactions will
not lead to oxidation of the functional moiety responsible for the
parent structures' antibactenal potency. However they may indirectly
influence the parent molecules' antibactenal activity either positively or
negatively, by altenng the parent molecule's bioavailability
11.1.1.3 Fluoroquinolones
Fluoroquinolones are believed to denve their biochemical activity from
several specific hydrogen-bonding and charge interactions with relaxed
bactenal DNA in the presence of DNA gyrase enzyme (Figure S4) (8).
According to the accepted model for fluoroqurnolone-DNA binding,
the charactenstic qumolone moiety is responsible for these interactions
(Figure S4). Oxidation of this moiety by O3 (as expected on the basis
of relatively fast reaction kinetics measured for the structurally-similar
substrate uracil-6-carboxylic, or isoorotic, acid (9J) should therefore lead
to a reduction or elimination of fluoroquinolones' antibactenal
potencies. However, the heterocyclic substituent groups (typically
192 Appendix II
piperazme denvatives, as for CF and EF — shown m Table 1 withm the
mam text) attached to many fluoroquinolones' qumolone moieties do
not appear to be essential to fluoroquinolone antibactenal activity For
example, first-generation qumolones such as nalidixic acid lack the
heterocycle, but still exhibit considerable antibactenal potency (10).
Thus, oxidation of the N(4) amme — expected on the basis of
secondary and tertiary amines' generally high reactivities toward
O3 (2,3) — may not contnbute to a significant reduction m CF's
antibactenal activity
II. 1.1.4 Dihydrofolate Reductase (DHFRJ Inhibitors
DHFR inhibitors (e.g., tnmethopnm, TMP — shown m Table 1 withm
the mam text) inhibit bactenal folate synthesis via competition with
dihydrofolate for DHFR enzyme (hence TMP's common role as a
synergist to the sulfonamide, SMX) (Figure S5) (5). The
diammopynmidme structure represents the active portion of these
antibactenal molecules, where the protonated nitrogen atom, N(l), of
the heterocyclic rmg (Figure S5) participates m charge interactions with
DHFR (11). Oxidation of the 2,4-diammopynmidme structure —
anticipated on the basis of pynmidme structures' generally high
reactivity toward O3 (9) — is therefore likely to yield a reduction m the
parent structure's antibactenal potency. TMP's 3,4,5-tnmethoxytolyl
moiety will also likely react relatively rapidly with O3 (12). As m the case
of sulfonamide R-substituent oxidation, this latter reaction may
indirectly influence the parent molecule's antibactenal activity by
altering its bioavailability
II.1.1.5 Lincosamides
Lmcosamides mteract with bactena via hydrogen-bondmg to specific
nucleotides m bactenal 23S rRNA, leadmg to inhibition of the bactena
cells' ability to synthesize proteins (Figure S6) (1). None of the
Oxidation of antibactenals: supporting information 193
functional moieties directly responsible for lmcos amide antibactenal
activity are expected to react appreciably with O3. However, O3 is likely
to oxidize the lmcosamide thioether moiety to its sulfoxide
denvative (13,14). In the case of lmcomycm (LM — shown m Table 1
withm the mam text), this reaction should lead to LM-sulfoxide, which
is known to possess significantly lower antibactenal potency than the
parent structure (15) — presumably due to interruption of requisite
mtermolecular hydrogen bondmg patterns at ad|acent hydroxyl and
methyl groups by thioether oxidation (Figure S6). It is unclear whether
oxidation of LM's pyrrolidine moiety would produce such an effect,
smce the pyrrolidine moiety is spatially and electronically isolated from
most of LM's biochemically-relevant functional moieties (Figure S6).
II. 1.1.6 ß-lactams
/^-lactams (including the ma|or penicillin and cephalosporin sub-classes)
denve their antibactenal activities from the fused /^-lactam rmg system,
which operates via sequestration of bactenal peptidoglycan
transpeptidase enzyme, leadmg to disruption of bactenal cell wall
synthesis (Figure S7) (5,16). The /^-lactam nng itself is unlikely to be
reactive toward O3. However, O3 can be expected to react with
yö-lactams' charactenstic thioether moieties (see PG and CP — Table 1 m
the mam text). Oxidation of the thioether by O3 is known to lead to
high yields (>95%) of many yö-lactams' R- and ^-sulfoxide enantiomers
(m R:S ratios ranging from 1:4 to 24:1) (13). Although .^-sulfoxide
analogues of /^-lactams exhibit negligible antibactenal activities, their
R-sulfoxide analogues are still quite potent (17,18), suggesting that
oxidation of the thioether by O3 may not be sufficient to eliminate the
parent /^-lactam compounds' antibactenal activities. Oxidation of the
double bond present m cephalosponn structures (e.g., cephalexin, CP —
shown m Table 1 withm the mam text) — most likely leading either to
ozonolytic cleavage or epoxidation (4,19) — may also be insufficient to
194 Appendix II
eliminate the parent structure's antibactenal properties, since the
/^-lactam rmg would not Hkely be disrupted by such a reaction.
Similarly, oxidation of CP's primary amine (Table 1) is not expected to
lead to significant reduction of the parent structure's antibactenal
activity
11.1.1.7 Tetracyclines
The antibactenal activities of tetracycline antibactenals appear to be
denved pnmanly from direct or indirect (metal cation-mediated)
electrostatic charge interactions between the keto and hydroxyl oxygens
on the underside of the tetracyclme molecule and the oxygens of
vanous mternucleotide phosphodiester lmkages on the 16S rRNA helix
contained withm the 30S subumt of the bactenal nbosome (Figure S8)
(20). Oxidation of the tetracyclic system by O3 — expected on the basis
of its activated unsaturated and aromatic character (2,4,21,22) — should
lead to reduction of the parent molecules' antibactenal properties, smce
such reactions would likely lead to extensive modification of the
relevant aromatic and olefinic moieties (4,22). However, oxidation of a
the tertiary amine moiety appears less likely to yield a significant
reduction m antibactenal potency of the parent molecule, since this
tnalkylamme moiety is spatially isolated from most biochemically-
relevant tetracycline functional moieties and not essential to the
tetracyclines' modes of action (Figure S8)
11.1.1.8 Glycopeptides
Glycopeptides (e.g., vancomycin, VM — shown m Table 1 withm the
mam text) are believed to denve their antibactenal activity from
sequestration of specific subumts of the peptidoglycan used m cell wall
synthesis (Figure S9) (23). Although none of the amide moieties
directly mvolved m the hydrogen-bondmg responsible for these
interactions should be appreciably reactive toward O3 (24), several of
Oxidation of antibacterials: supporting information 195
the ad|acent aromatic moieties (Table 1) should be highly reactive
(2,12). One would expect that ozonation of VM's phenol or
tnmethoxybenzyl moieties by O3 should lead to nng cleavage, oxidation
to qumone structures, or formation of oxyl radicals (the latter of which
could foreseeably cross-link with proximal components of the VM
structure) (22,25). Each of these structural modifications would likely
lead at least to impairment, if not elimination of VM's antibactenal
activity, via interruption of the specific VM stereochemistry necessary
for binding of bactenal peptidoglycan (Figure S9). Oxidation of the
secondary N-methylleucme moiety by O3 (presumably leading to
hydroxylation or dealkylation of the amine (26)) may also lead to
disruption of hydrogen bondmg at the ad|acent amide (Figure S9).
However, oxidation of the primary vancosamme will not likely lead to
sufficient disruption of stereochemistry to substantially dimmish VM's
antibactenal potency, as a consequence of the vancosamme moiety's
isolation from VM's biochemically-relevant amide moieties (Figure S9).
II.1.1.9 Aminoglycosides
Aminoglycosides — which are inhibitors of bactenal protein synthesis —
denve much of their antimicrobial activity from charge mteractions with
nucleobase and phosphodiester functional groups m bactenal rRNA (27).
One proposed bmdmg scheme — developed with streptomycin as a
model — suggests that these mteractions mvolve vanous hydroxyl and
amme groups withm the typical ammoglycoside structure (Figure S10).
Aminoglycosides also contribute to disruption of cell walls; an attribute
apparently denved m large part from their ability to displace Mg^+ and
Ca2+ from outer membrane-associated lipopolysacchande bndges (28).
Oxidation of an aminoglycoside's primary amme moieties by O3 (e.g., to
nitro groups, ammoxides, or via deammation (14)) would presumably
prevent the bactenal rRNA hydrogen bondmg and catiomc charge-
interactions from which ammoglycoside antibactenal activity is largely
196 Appendix II
denved (Figure SIO), m turn leadmg to reduction or elimination of the
parent compounds' antibactenal potencies.
11.1.2 S2. Chemical reagents
Commercially-available antibactenal substrates and substructure model
substrates were purchased from Sigma-Aldnch, with the exception of
cephalexin (CP) hydrate — which was purchased from MP Biomedicals.
Azithromycin (AZ) dihydrate was a gift from Pfizer, Inc. 2,4-diammo-
5-methyl pynmidme (DAMP) — produced by Daniels Fine Chemicals,
Ltd., and 4-ammophenyl methyl sulfone — produced by Sigma-Aldnch,
were gifts from Professor Chmg-Hua Huang, Georgia Institute of
Technology, Atlanta, GA. N(4)-acetyl-sulfamethoxazole (ASMX) was
synthesized as descnbed by Gobel et al. (29). All substrates were of
95% punty or higher, with the exception of ASMX — which was~ 70%
pure. Other chemicals (e.g., buffers, H2O2, reducing agents, etc.) were
of reagent grade quality or better. O3 stock solutions were produced as
descnbed previously (30), and standardized accordmg to direct O3
absorbance at X = 258 nm (usmg 6 S 3000 M 'cm 1). Stock solutions of
antibactenals and substructure models were prepared m Nanopure or
M1II1-Q water. Acetone and acetonitnle — «OH scavengers which should
not accelerate radical-chain O3 decay m the aqueous systems typical of
this study (31-33) — were used to facilitate preparation of several poorly
soluble reactant stock solutions for O3 rate constant determinations.
Acetone and acetonitnle concentrations never exceeded 1 % by
volume, so co-solvent effects should have been negligible (34).
Solutions used for determination of «OH rate constants and for
municipal wastewater expenments contained no co-solvents.
Oxidation of antibacterials: supporting information 197
11.1.3 S3. Measurement of apparent second-order rate
constants for O3 and »OH reactions
11.1.3.1 Method I (O3 kinetics]
UsedforRX, AZ, AM, CH, CHU, DMCH, EPÇ andMP (Tables 1 and2
within the main text). Rate constants for model substrates which react
relatively slowly with O3, but do not absorb UV radiation appreciably at
X = 258 nm were measured by directly following the consumption of
O3 at this wavelength. These expenments were conducted under
conditions of excess substrate, where [substrate]o:[03]o was at least 10:1
(typically 20:1 or greater) m all cases. Pseudo-first-order rate constants
determined from plots of ln([03]/[03]o) vs. time were linear (r2 > 0.98)
m all cases. A stopped-flow spectrophotometry system was constructed
m house to permit measurement of pseudo-first-order rate constants
k'03,obs,M up to approximately 2.3 s1. Solutions of substrate and O3 were
prepared at 100 to 200 uM and 10 to 20 uM (to yield 50 to 100 uM and
5 to 10 uM after 1:1 mixing), respectively, m reagent water buffered to
the desired pH with approximately 0.01 M phosphate. These solutions
were m|ected at flow rates ranging from 25-35 mL/mm from two self-
contained Dosimat syringe pumps (Metrohm, Switzerland), through
1.5 mm I.D. PEEK tubing, and into the mlet ports of a 60° mner angle
PEEK mixing tee. The outlet of the mixmg tee was connected with a
~7 cm length of 0.5 mm I.D. FEP tubing to a 5 cm flow-through UV-
spectrophotometry cell with a cylmdncal 3 mm I.D. optical path. The
effluent line of the stopped-flow system was connected via PEEK
tubing to a 25-mL luer-lock gas-tight synnge clamped on to a lab-stand
with the needle-side facmg down, and the outside end of the piston
several centimeters below a fixed metal plate. Effluent from the
spectrophotometer cell flowed mto the needle-end of the synnge until
the piston contacted the metal plate — resulting m abrupt stoppage of
system flow. Reactions were monitored after stoppage by following
198 Appendix II
decay of O3 at X = 258 nm. The effective dead-time for this system
was calculated to be 0.08 + 0.21 = 0.29 ~ 0.3 s (mmimum mstrument
resolution of the spectrophotometer was 0.1 s). On the basis of this
dead-time, the limit of accurate measurement was estimated to be that
for a reaction with ti/2 of approximately 0.3 s, or k'o3,obs ~ 2.3 s '. This
translated to a practical quantification limit of k"o3,obs ~ 4.6 x 104 M 's ',
for [substratejo = 50 uM. The stopped-flow apparatus was calibrated by
companson of the second-order rate constant determmed for the
neutral species of RX (k"o3,neutadiRx = 1-2 (+ 0.1) x 107 M 's ') with that
reported previously (k"o3,neutndiRx = 1-0 (+ 0.1) x 107 M's ' (35)), after
calculation of the latter value for pK^ = 9.2 (36) (a pKa of 8.8 was used
m the pnor study).
11.1.3.2 Method II (O3 kinetics]
Used for ASMX, CF (pH 3-6), ET (pH 3-5.5), PG, DMI, and FTU
(Tables 1 and 2 within the main text). Rate constants for slowly-reacting
substrates which absorb strongly at wavelengths close to 258 nm could
not be determmed by direct measurement of O3 decay. These rate
constants were instead determined under pseudo-first-order conditions
of excess O3 ([O3]o:[substratejo > 20) by foUowmgloss of substrate via
HPLC with UV or fluorescence detection. All HPLC analyses were
performed on either a Hewlett-Packard 1050 HPLC system equipped
with a Supelco Discovery RP Amide C16 column (3 mm x 250 mm, 5
uM), fluorescence detector (FLD), and single-wavelength UV detector,
or an Agilent 1100 HPLC system equipped with the same column and a
vanable wavelength UV diode-array detector. Separations were
performed with acetonitnle and 0.05 M H3PO4 (ad|usted to pH 2.2
with NaOH) as mobile phases, using isocratic or gradient methods as
required for the analyte(s) of mterest. UV detection was performed at
wavelengths from 205 to 280 nm, depending on the analyte (limits of
quantification S 0.05 uM). Fluorescence detection was performed at
Oxidation of antibacterials: supporting information 199
Xex = 278 nm and Xem = 445 nm for [CFJo and [EFJo < 1 uM, and at
Xex = 278 nm and Xem = 360 nm for [FLUJo < 1 uM (limits of
quantification S 0.01 uM). Standard deviations for each measurement
vaned from +1-5%.
Reactions were initiated by m|ectmg — under constant stirring — 10-20-
fold excess concentration of O3 mto 1-5 uM solutions of substrate
contained withm 100-mL amber, borosilicate bottles fitted with screw-
cap piston-type dispensers (37), and thermostatted by a recirculating
water bath. 1-mL samples were subsequently dispensed at pre¬
determined time mtervals mto amber, borosilicate HPLC vials containing
25 uM of 10 mM cmnamic acid to quench residual O3, and subsequently
transferred to HPLC for analysis of residual substrate, O3 (as
benzaldehyde (38)), and/>-chlorobenzoic acid (pŒ>A — used as an internal
standard used to correct for sample dispensation maccuracies). Pseudo-
first-order rate constants determmed from plots of ln([M]:[M]o) (where
M represents the model substrate) vs. time were Imear (r2 > 0.98) m all
cases. All expenments were conducted m duplicate.
11.1.3.3 Method III (O3 kinetics]
Usedfor CP, TET, DAMP, and MBDCH (Tables 1 and 2 within the main
text). A number of model substrates mcluded m this study do not
significantly absorb UV radiation at wavelengths above 200 nm, are
unstable m aqueous solution of extended penods of time, or are
difficult to analyze, but react too rapidly with O3 to follow by available
manual methods. In each of these cases O3 reaction rate constants were
determined by application of a competition kmetics method requinng
the measurement of only a single endpomt P, which is typically the
formation of a product resultmg from oxidation of either the model
substrate, M, or competitor substrate, C (12). Expenments were
conducted by addmg a fixed dosage of O3 to rapidly-stirred, 20-mL
200 Appendix II
volumes of ten different buffered solutions containing vanous ratios of
[M]o:[C]o (both m at least 10-fold molar excess of O3). Cmnamic acid
(k'oveutod = 5 x 104 M's1 and k'o^on = 3.8 x 105 M's1 (38),
pK, = 4.4 (39)) or buten-3-ol (k"03 = 7.9 x 104 M 's 1 (4J) were used as
competitor substrates m these expenments. Yields of benzaldehyde —
measured by HPLC-UV — or formaldehyde — determmed
spectrophotometncally at X = 412 nm as diacetyldihydrolutidme (40) —
were selected as respective measurement endpomts, P. Model substrate
rate constants were evaluated using eq SI,
Fjabsence=1|
kO„app,clC\o^
l"Jpresence ^o3,app,M V^Jo
where [Pjabsence represents the measured endpomt yield m the absence
of competitor substrate (obtamed from duplicate C controls), and
[P]Presence represents endpomt yield m the presence of varymg doses of
competitor substrate. k"03jilpBc was determmed from the slopes of plots
Of ([P]abSence/[P]presence "1) VS. [q0/[M]0.
11.1.3.4 Method IV (O3 kinetics]
Used for TYL, SMX, TMP, CF (pH 6.5-8), EF (pH 6-8), VM, APMS,
and TMT (Tables 1 and 2 withm the mam text). A competition kmetics
method requinng the measurement of two-endpomts was used to
determine k"o3,aPP for substrates with k"o3,aPP> 5 x 103 M 's ' that absorb
UV radiation strongly at wavelengths above 200 nm (35). In this
approach, a different O3 dose was added to each of ten rapidly-stirred,
20-mL volumes of a buffered solution containing a fixed ratio of
[M]0:[C]o. The two endpomts — residual reference substrate (cmnamic
acid) and competitor substrate concentrations remaining after O3
addition — were measured by HPLC-UV Model substrate rate constants
were evaluated according to eq S2,
Oxidation of antibacterials: supporting information 201
In[C])_J[M]
In
[C\) [M03 ,app,C
03,app,M
(S2)
where k"o3,iPBc could be determmed from the slope of a plot of
ln([q/[q0) vs. ln([M]/[M]o). [M]0 and [CJowere taken from analyses of
duplicate O3 controls mcluded m each expenment.
11.1.3.5 Method V ('OH kinetics]
Used for RX, AZ, ASMX, TMP, LM, PG, CP, and VM. «OH rate
constants were determmed for photo-stable model substrates via
application of the same competition kmetics approach as utilized for
Method IV, except that eq S2 was modified to eq S3.
(
In[C])_J[PCBA]
=ln
A
[C\ [\pCBA\•OH,app,C (S3)
•OH,app,pCBA
Hydroxyl radicals were generated by m situ UV-photolysis of H2O2 m
solutions containing the reference substrate p-chlorobenzoic acid
(pCBA, with k".oH,a.pp,pCBA = 5.0 x 109 M h 1 (41)), and competitor
substrate, C, according to a previously descnbed procedure (35). Bnefly,
reaction solutions buffered at pH 7 with approximately 0.01 M
phosphate were dosed with 2 mM H2O2, and fixed [C]o:[pCBA]o. These
solutions were irradiated with a 500-W medium pressure lamp (X < 308
nm screened by a UV-cutoff filter) for repeated ten-mmute intervals up
to two-hour total irradiation penods. Samples were withdrawn from
these solutions between each irradiation penod for residual substrate
analysis by HPLC-UV RX, AZ, and LM were detected at X = 205 nm.
Expenments included a minimum of 10 different irradiation times.
H2O2 controls were included m all expenments.
202 Appendix II
11.1.3.6 Method VI ('OH kinetics]
Usedfor TET, TYT, CF, andEF. These substrates were moderately to highly
photo-labile m the presence of UV radiation passing the 308 nm cutoff
filter used m Method V »OH rate constants were determined for these
substrates accordmg to the same competition kmetics procedure used m
Method V (i.e., via eq S3), but usmg y-radiolysis (42) to generate »OH m
solutions contammg each of the model substrates and/>CBA Expenments
mcluded at least eight different irradiation times (»OH exposures).
11.1.3.7 Method VII ('OH kinetics]
UsedfiorAM. The »OH rate constant for AM was determined accordmg to
Method V, except that AM was denvatized with 9-fluoroenylmethyl
chloroformate (FMOQ (43), to permit detection by UV One-mL samples
(contammg starting concentrations of 5 uM AM, 5 \xMpCBA, and 2 mM
H2O2) taken from the photo-reaction system descnbed above were dosed
mto amber, borosilicate HPLC vials containing 25 mM of sodium pyruvate
to quench residual H2O2 (44) — which appeared to interfere with the
denvatization process — and 25 mM of NaOH (to prevent losses of AM to
adsorption onto glass surfaces pnor to denvatization (45) and to accelerate
the reaction between pyruvate and H2O2 (44,46)). 500 uL of these samples
were subsequently transferred to separate HPLC vials containing 500 uL
of 2.5 mM FMOC m acetonitnle and 200 uL of 0.8 M bone acid (to
maintain a solution pH of ~ 9) for denvatization (43). The remainder of
the undenvatized samples were used for analysis of residual ^CBA by
HPLC-UV (X = 205 nm). Denvatized AM was measured by HPLC-UV
(A, = 200 nm) a minimum of one hour after initiation of the denvatization
reaction (15 minute reaction times are reported to be sufficient for
denvatization of gentamicm, another aminoglycoside (43)).
AM concentrations m dark controls containing H2O2 were stable over the
135-mmute experimental penod, indicating that adsorption of AM to
reactor tube surfaces was negligible dunng the penod of each expenment.
Oxidation of antibacterials: supporting information 203
11.1.4 S4. Wastewater matrix experiments.
A 10-L grab sample of secondary wastewater effluent was obtained
from the Kloten-Opfikon wastewater treatment facility near Zunch,
Switzerland. This sample was transported to the laboratory in a
polypropylene carboy withm several hours of sampling and vacuum-
filtered with a 0.45 urn cellulose-nitrate membrane pnor to storage
at 4°C. Sample alkalinity and DOC were 3.5 mM as HCO3 and
5.3 mgC/L, respectively Native sample pH was 7.7. Kloten-Opfikon
secondary wastewater effluent is known to contain - on average - less
than 1 ug/L of RX, AZ, SMX, ASMX, and TMP (47,48). The
concentration of CF m the influent to this plant has been reported as
< 1 ug/L on average, and should be significantly lower after secondary
treatment. These levels are negligible relative to the concentrations
spiked to Kloten-Opfikon sample for wastewater ozonation
expenments, which were 1 uM for each substrate (translating to a
mmimum concentration of 253 ug/L, for - SMX - the substrate with
the lowest molecular weight).
Only the reaction between O3 and ASMX was slow enough to permit
direct monitoring of ASMX loss dunng ozonation of the wastewater
sample. 63 mM (3 mg/L) of O3 was added under constant, rapid
stirring to a 20-mL volume of Kloten-Opfikon wastewater contammg
1 uM of ASMX and 1 uM of pCBA. Samples were taken every 10 s
until O3 was completely depleted, for subsequent analysis of residual
ASMX and pCBA by HPLC-UV
For all other substrates, vanous doses of O3 (ranging from 5 uM
(0.25 mg/L) to 104 uM (5 mg/L)) were added under constant, rapid
stirnng to 20-mL volumes of Kloten-Opfikon wastewater — each
contammg 1 uM of the model substrate of mterest and 1 uM of ^CBA.
Reactions were allowed to proceed until complete O3 depletion (except
m the case of solutions dosed with 104 uM O3 doses, which were
204 Appendix II
quenched with buten-3-ol after ~ 90 s of reaction to remove O3
residual). One-mL samples were then taken from each reaction solution
and transferred to amber, borosilicate HPLC vials for direct HPLC-UV
analysis of residual model substrate andpCBA concentrations.
Each expenment was conducted m duplicate. In addition, duplicate O3
controls were included with each set of samples to venfy substrate
stability m the wastewater matrix. Analyte recovenes were calculated
from measurements obtamed for control samples of distilled water
dosed with 1 uM concentrations of each substrate, and found to be
100 + 10% for all analytes except TET, for which a recovery of 117%
was recorded. All expenments were conducted at T = 20 + 0.5 °C.
11.1.5 S5. Estimation of transformation efficiencies and
f.oHMfor RX, AZ, AM, and LM
Transformation of a substrate (M) is governed by eq S4 (49) dunng
ozonation of a real water, where the two integral terms on the nght
hand-side represent the O3 and «OH exposures governing its
conversion (M/Mo) for a reaction time, t, where [*OH] = f(t) and
[O3] = g(t).
In
Mo-^app4[03]^-£.oH^MJ[-OH>/f (S4)
The «OH exposure term, J'OHdt can be calculated by usmg an
03-recalcitrant compound such as pCBA as a probe to measure m situ
•OH exposures, according to eq S5 (49,50). Theoretically, such an
}[«OH>/f1
^•OH.app.^CBA
-In
[pCba\(S5)
Oxidation of antibacterials: supporting information 205
approach should also be applicable to determining relative O3
exposures for substrates with k"03japp > 103 M h 1 dunng ozonation
processes; i.e., one should be able to determme JOjdt for a certain
probe compound and use this value to estimate ln([M]/[M]o) for other
substrates exposed to the same reaction conditions. However, unlike
•OH exposure, O3 exposure can only be calculated indirectly from eqs
S4 and S5 (via eq S6), since no •OH-refractory, 03-susceptible probe
compound is available to provide an m situ measurement of this value.
TF
U.OHapp,CBA l^CBA]0J IKJJ/ °>U
The applicability of this approach was evaluated by calculating
ln([M]/[M]o) (at O3 dosages of 1 and 1.5 mg/L) for each of the nme
substrates TYL, SMX, TMP, CF, EF, CP, PG, TET, and VM, by using
each of the JOjdt terms determmed with eq S6 for the other eight
substrates. J'OHdt was calculated for each substrate accordmg to eq S5.
As shown m Figure S13, this procedure yielded good estimâtes of
transformation efficiencies, ln([M]/[M]o), as long as JOjdt terms were
selected from the JOjdt value measured for the substrate with the most
similar k"03japp at the wastewater pH of 7.7 (e.g., PG transformation
was estimated on the basis of CF transformation, EF transformation
from TMP transformation, and TET transformation from TYL
transformation). In contrast, significant discrepancies were observed
when estimates were based on JOjdt obtamed from substrates with
markedly different magnitudes of k"o3,aPP (i.e., differing by more than one
order of magnitude from each other). Estimates of ln([M]/[M]o) for RX,
AZ, AM, and LM were obtained by usmg the JOjdt terms for TMP,
SMX, CF, and TYL, respectively These estimates were m turn used to
calculate £oh,m by eq 6 m the mam text (see Figure 4 m the mam text).
206 Appendix II
The discrepancies noted above — for ln([M]/[M]o) estimates based on
companson of substrate pairs with dissimilar k"o3,aPP values — may have
been a consequence of incomplete mixing on the time-scales of initial
O3 consumption m each reaction system. Smce initial consumption of
O3 by wastewater matrixes such as that utilized here is typically
extremely rapid (ti/2 on the order of 0.01 to 0.05 s) (51), such a mixing
effect may have resulted m disproportionate O3 consumption by locally
abundant sample matnx constituents (relative to very small local
concentrations of the given antibactenal substrate) withm the vicinity
of O3 m|ection. This could have resulted m lower observed
transformation of substrates than predicted on the basis of their O3
rate constants alone. Estimation errors due to such an effect would be
more pronounced for substrate pairs exhibiting large differences m
k"o3,aPP (i.e., > 102), as indicated above
Oxidation of antibacterials: supporting information 207
11.2.1
Tables
SI. pKa values and corresponding source
references for each antibacterial substrate
Structural Class Substrate* pJTal PJT»2 PJT»3 pJT»< pjr* pJTaS
Macrolide RX
AZ
TYL
9 2 (36)
8 7 (52)
7 7 (52)
9 5 (52)
Sulfonamide SMX
ASMX
1 7 (53)
5 5 (54)
5 6 (53)
Fluoroquinolone CF
EF
6 2 (55)
6 1 (56)
8 8 (55)
7 7 (56)
DHFR Inhibitor TMP 32
(36,57,58)
7 1 (58)
Lmcosamide
/^-lactam
LM
PG
CP
7 8 (36)
2 7 (59)
2 5 (60) 7 1 (so;
Tetracycline TET 3 3 (61) 7 7 (Sij 9 7 (eij
Glycopeptide VM 2 9(62,63) 72
(62,63)
86
(62,63)
9 6 (62) 10 5 (62) 11 1 (62)
Aminoglycoside AM 6 7 (64) 8 4 f»<J 8 4 f»<J 9 7 f«J
*For full names, see Table 1 in the main text
11.2.2 S2. pKa values and corresponding source
references for each substructure model substrate
Substrate" Pjfal
DMCH 10 7 (65)
MP 10 2(66)
APMS 1 5 (67)
DMI NA
EPC 8 3 (68)
FLU 6 5 (69)
DAMP 3 2b
TMT NA
MBDCH 3 5 (66)
CH 10 6(65)
CHM 10 3(66)
pATa2
NA — Not applicable. ^For full names, see
Table 2 in the main text. ^pi*Q values
assumed to approximate those for TMP,since DAMP is identical to TMP's 2,4-
chanrino-5-memylpyrimdine moiety.
208 Appendix II
11.2.3 S3. Second-order rate constants (AAV1) for reactions
of O3 with substructure model substrates
Substrate"
(Rateconstant
measurement
methods ) Diprotonated
Species
Monoprotonated Deprotonated
DMCH (I) NA <id 3 7 (± 0 1) x 106
MP (I) NA <id 2 0 (± 0 1) x 106
APMS (IV) NA ND 4 7 (± 0 1) x 10"
DMI (II) NA NA 5 4 (± 0 3) x 101
EPC (I) NA <ld 1 1 (± 0 1) x 106
FLU (II) NA 1 2(± 0 7) 1 8 (± 0 1) x 103
DAMP (III) 5 0 (± 12) x 102 2 9(± 1 3)x 103 1 3 (± 0 2) x 106
TMT (IV) NA NA 2 8(±0 l)x 10s
MBDCH (III) NA ND 1 4 (± 0 4) x 106
CH(I) NA <ld 4 9 (± 0 2) x 10"
CHM (I) NA <ld 7 1 (± 0 2) x 10"
NA — Not applicable, ND — Not determined ^For full names, see Table 2 in the
main text. ^Described in Text S3. O3 reaction rate constants were measured at
T = 20 i 0.5° G 'These rate constants were assumed to be negligible, on the
basis of prior observations for protonated amine reaction centers (2,24).
Oxidation of antibactenals: supporting information 209
Figures
^
^ </ .^ .crP x#
^
3 MIC -S Aureus
3 MICa-£ faecalis
i MICa-£ coll
I Hospital Sewagel Raw Sewage and
Primary Effluent
] Final Effluent
] Surface Water
Maximum single-compound* concentrations of various antibacterial
classes detected in municipal wastewater systems and surface waters, in
the context of minimum reported clinical MIC values for sensitive
bacterial reference strains Data for fluoroquinolones (70-76),sulfonamides (48,72-78), DHFR (dihydrofolate reductase) inhibitors
(72,73,75-77,79,80), tetracyclines (72,73,80), /^lactams (81), macrolides
(29,48,73,74,76,82), aminoglycosides (83), and lincosamides (73) was
obtained from various environmental analytical studies ^MIC —
minimal inhibitory concentration, or minimal concentration resulting in
a measurable reduction in bacterial growth relative to an antibacterial
blank MICs listed for each class correspond (in order from left to
right) to reported values for ciprofloxacin, sulfamethoxazole,
trimethoprim, tetracycline, penicillin, erythromycin, gentamicin, and
clindamycin, respectively (10), ^MIC values not reported for E. coll
*More than one compound from a given antibacterial class may be
present in the same municipal wastewater
210 Appendix II
.3.2 S2
Roxithromycin (RX) N'
O^ N
U2609
_OH H2N^N
N^N~A2040
Biochemical model for mechanism of macrolide antibacterial activity
(adapted with permission from Schlunzen et al (1))
II.3.3 S3
7 8-dihydroptenn
pyrophosphate
®®
0=( ,NHN-
NH,
Dihydropteroatesynthase
Sulfonamide (e g SMX)*
pABA
Dihydropteroatesynthase
HN
H2N"St"NVnrOf*.
HN
Dihydropteroate
N^ -N-fVCA, iL J h N=/ bH
Biochemical mechanism of sulfonamide antibacterial activity (5) *R
represents an aromatic substituent that vanes, depending on the
particular sulfonamide compound
Oxidation of antibactenals: supporting information 211
.3.4 S4
Hydrogen bondingto nucleotides
DNA strand
Biochemical model for mechanism of fluoroquinolone antibacterial
activity (adapted with permission from Shen et al (8))
212 Appendix II
.3.5 S5
COOH
HOOC-
CX,NH
O^J N
hnXNH2
7 8-dihydrofolate
DHFR Inhibitor (eg TMP)* NH2
Metabolic dead-end
N3 =^nXi
H2N N
Dihydrofolate reductase
9 H
5 6 7 8-tetrahydrofolate
H2N N N
H
P COOH
N •/ COOH
H
Biochemical mechanism of dihydrofolate reductase (DHFR) inhibitor
antibacterial activity (5) *R represents an aromatic substituent that
varies, depending on the particular DHFR inhibitor
.3.6 S6
G2505EC O^qh N A2058EC
II
(U2590Dr)NH C2611EC
(A2040Dr)G2057EC
Lincomycin N NH2
Biochemical model for mechanism of lincosamide antibacterial activity
(adapted with permission from Schlunzen et al (1))
Oxidation of antibactenals: supporting information 213
.3.7 S7
O NH
r v v xsHO 7 ' HO /x
Enz + p-lactam Non-covalent
enzyme complex
^ 9hA>NH Very slow
\ HN-[.'H- HO'
O^ + Enz-OH
O, .NH
HO AHKH"H
IAcylated-enzyme |_|q
Fragmentation products
Biochemical mechanism of /^-lactam antibacterial activity (depicted for
penicillin G) (5,16).
.3.8 S8
HO, ,<
OH O OH O O O
Tetracycline G966° p °
G1053 o
O P O'
H2N è$
C1054
C1195
G1198
Biochemical model for primary* mechanism of tetracyclineantibacterial activity (adapted with permission from
Broderson et al (20)) *Tetracychne binding at a secondary site within
the bacterial ribosome is beheved to involve many of the same
functional moieties as binding at the primary site (20)
214 Appendix II
.3.9 S?
OH"
NH,
Oh?-
HO^^OHO^l0^0 Vancomycin
Ol X XX 9
Biochemical model for mechanism of glycopeptide antibacterial activity
(adapted with permission from Williams and Bardsley (1999) (23))
Oxidation of antibactenals: supporting information 215
.3.10 S10
NH2HN^
HQ. nhStreptomycin
HN-X' ) OH
C1490
h2nx yy,c
-o
i2in \\ / vNhO OH
/
HN
HO
O
HOHO
A914 O O HN^ Lys45
qP (S12)
C526
U14G527
Biochemical model for mechanism of aminoglycoside antibacterial
activity (adapted with permission from Carter et al (27))
216 Appendix II
.3.11 Sil
0 001 100
»1/2 (s)
Calculated ti/2 values for the apparent transformation of model
antibacterial substrates by O3, in companson to correspondingestimated half-lives for reaction of O3 with the targeted functional
moieties (Table 1 in the main text) associated with each substrate's
biochemical activity at 20(±0 5) °C, pH 7, and [O3] = 42 uM (2 mg/L)^Either O3 does not appear to react directiy with biochemically-active
target moieties (PG, CP), or ti/2 for this reaction could not be
determined (ASMX) ^ti/2 targetestimates for CF and EF are based on
the ti/2 value determined for FLU and ti/2 targetfor LM based on ti/2
determined for cationic LM, as discussed in the main text
Oxidation of antibacterials: supporting information 217
.3.12 SI 2
i o<5~
^ 06 -
CO
O Measured pCBA loss
• Measured ASMX loss
O Measured 03 loss
Calculated f.nH
0 10 20 30 40 50 60
time (s)
Transformation of ASMX during ozonation of Kloten-Opfikon wastewater
at 20(±0.5) °C, pH 7.7, [O3]o = 63 U.M (3 mg/L), and [substrate]0 = 1 U.M.
S represents each monitored substance — 03,^CBA, and ASMX.
.3.13 SI 3
57
• [O3]0 = 21 nM (1 mg/L)
O [03]„ = 31 nM (1 5 mg/L)
04 06
Measured [P]/[P]0
1 0
Correlations of predicted substrate transformation with measured values, for
data sets obtained at T = 20° C, pH 7.7, and [O3]0 = 1 and 1.5 mg/L (21 and
31 UM, respectively).
218 Appendix II
.3.14 SI 4
J"" 0 00
0 00 0 05 0 10 0 15 0 20
f.OHH(t) calculated from direct measurements (eq (3))
Comparison of f-oHM(x) values calculated from indirect determinations
of O3 exposure (by eq 6 in the main text) with those calculated from
direct measurements of O3 exposure (by eq 4 in the main text)Calculations were performed with data reported elsewhere (51) for
[carbamazepine]o = 1-2 uM, [O3] = 25-50 uM (1 2-2 4 mg/L), in
vanous municipal wastewater samples
Literature Cited
Schlunzen, F, Zanvach, R, Harms, J, Bashan, A, Tocilj, A, Albrecht, R,
Yonath, A, Franceschi, F, Structural basis for the interaction of antibiotics
with the peptidyl transferase centre in eubactena Nature 2001, 413, 814-821
Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water — II Dissociating organic
compounds Water Res. 1983, 17, 185-194
Muiioz, F, von Sonntag, C,The reactions of ozone with tertiary amines
including the complexing agents nitrilotnacetic acid (NTA) and
ethylenediaminetetraacetic acid (EDTA) in aqueous solution J. Chem. Soc,
Perkm Trans. 2 2000, 2029-2033
Dowideit, P, von Sonntag, C, Reaction of ozone with ethene and its
methyl- and chlorine-substituted denvatives in aqueous solution Environ.
Sa. Technol. 1998, 32, 1112-1119
Oxidation of antibactenals: supporting information 219
5 Walsh, C Antibiotics:Actions, Origins, Resistance, ASM Press Washington, DC, 2003
6 Spanggord, R J, Yao, D, Mill, T, Kinetics of amtnodimtrotoluene
oxidations with ozone and hydroxyl radical environmental Science &
Technology 2000, 34, 450-454
7 Pierpoint, A C, Hapeman, C J, Torrents, A ,
Linear free energy study of
ring-substituted aniline ozonation for developing treatment of aniline-
based pesticide wastes J. Agnc. Food Chem. 2001, 49, 3827-3832
8 Shen, L L, Mitscher, L A, Sharma, P N, O'Donnell, T J, Chu, D W T,
Cooper, C S, Rosen, T, Pernet, A G
,Mechanism of inhibition of DNA
gyrase by qumolone antibactenals a cooperative drug-DNA bindingmodel Biochemistry 1989, 28, 3886-3894
9 Theruvathu, J A , Flyunt, R , Aravindakumar, C T, von Sonntag, C ,Rate
constants of ozone reactions with DNA, its constituents, and related
compounds J. Chem. Soc, Perkm Trans. 2 2001, 269-274
10 Wiedemann, B, Grimm, H In Antibiotics in Laboratory Medicine, Lonan, V,
Ed, Williams and Wilkins Baltimore, MD, 1996, pp 900-1168
11 Cocco, L, Roth, B, Temple, C,Jr, Montgomery, J A,London,R E,Blakley,RL, Protonated state of methotrexate, trimethoprim, and pyrimethamine bound
to dihydrofolate reductase Arch. Biochem. Biophys. 1983, 226, 567-577
12 Muhoz, F, von Sonntag, C, Determination of fast ozone reactions in
aqueous solution by competition kinetics J. Chem. Soc, Perkm Trans. 2
2000, 2, 661-664
13 Spry, D O, Oxidation of penicillin and dihydrocephalosporin derivatives
with ozone J. Org. Chem. 1972, 37, 793-795
14 Bailey, P S Ozonation in organic chemistry. II. Non-olefinic compounds, Academic
Press New York, NY, 1982
15 Sztaricskal, F, Dinya, Z, Batta, G, Mocsari, A, Chemical synthesis and
structural study of lincomycin sulfoxides and a sulfone The Journal ofAntibiotics 1997, 50, 866-873
16 Livermore, D M, Williams, J D In Antibiotics m Laboratory Medicine, 4 ed,
Lonan, V, Ed, Williams and Wilkins Baltimore, MD, 1996, pp 502-578
17 de Körting, J J, Marx, A F, Poot, M M, Smid, P M
, Verwei], J In Recent
Advances m the Chemistry of ß-Lactam Antibiotics, Elks, J, Ed, The Chemical
Society Burlington House, London (England), 1977, Vol 28, pp 161-166
18 Torn, H, Asano, T, Matsumoto, N, Kato, K, Tsushima, S, Kakinuma,
A,Microbial oxidation of cephalosporins to cephalosporin sulfoxides
Agricultural and Biological Chemistry 1980, 44, 1431-1433
19 Bailey, P S Ozonation m organic chemistry. I. Olefinic compounds, Academic
Press New York, NY, 1978
20 Brodersen, D E, Clemons, W M , Jr, Carter, A P, Morgan-Warren, R J ,
Wimberly, B T, Ramaknshnan, V, The structural basis for the action of
220 Appendix II
the antibiotics tetracycline, pactamycin, and hygromycin B on the 30S
ribosomal subunit CelllOOO, 103
21 Hoigné, J, Bader, H ,Rate constants of reactions of ozone with organic
and inorganic compounds in water — I Non-dissociating organic
compounds Water Res. 1983, 17, 173-183
22 Mvula, E, von Sonntag, C, Ozonolysis of phenols in aqueous solution
Org. Biomol. Chem. 2003, /, 1749-1756
23 Williams, D H, Bardsley, B, The vancomycin group of antibiotics and
the fight against resistant bactena Angewandte Chemie-International Edition
1999,38, 1173-1193
24 Pryor, W A, Giamalva, D H, Church, D F, Kinetics of ozonation 2
Amino acids and model compounds in water and comparisons to rates in
nonpolar solvents J.Am. Chem. Soc. 1984, 106, 7094-7100
25 Ragnar, M , Eriksson, T, Reitberger, T, Brandt, P, A new mechanism in the
ozone reaction with lignin like structures Hol^forschung 1999, 53, 423-428
26 von Gunten, U, Ozonation of drinking water Part I Oxidation kinetics
and product formation Water Res. 2003, 37, 1443-1467
27 Carter, A P, Clemons, W M , Jr, Brodersen, D E,Morgan-Warren, R J ,
Wimberly B T, Ramakrishnan, V, Functional insights from the structure
of the 30S ribosomal subunit and its interactions with antibiotics Nature
2000, 407, 340-348
28 Kadurugamuwa, J L, Lam, J S, Bevendge, T J, Interaction of
gentamicin with the A band and B band lipopolysacchandes of
Pseudomonas aeruginosa and its possible lethal effect Antimicrob. AgentsChemother. 1993, 37, 715-721
29 Gobel, A, McArdell, C S, Suter, M J-F, Giger, W, Trace détermination of
macrohde and sulfonamide antimicrobials, a human sulfonamide metabolite,
and tnniethopnni in wastewater using liquid chromatography coupled to
electrospray tandem mass spectrometry Anal. Chem. 2004, 76, 4756-4764
30 Bader, H, Hoigné, J, Determination of ozone in water by the indigomethod Water Res. 1981, /5, 449-456
31 Hoigné, J, Bader, H ,The role of hydroxyl radical reactions in ozonation
processes in aqueous solutions Water Res. 1976, 10, 377-386
32 Staehelin, J , Hoigné, J , Decomposition of ozone in water in the presence
of organic solutes acting as promoters and inhibitors of radical chain
reactions Environ. Sa. Technol. 1985, 19, 1206-1213
33 Mossen, S, Neta, P, Meisel, D, The mechanism of cyanide release in the
radiolysis of acetonitnle Formation and decay of the cyanomethylperoxylradical Radiai. Phys. Chem. 1990, 36, 683-687
34 Schwarzenbach, R P, Gschwend, P M, Imboden, D M Environmental
Organic Chemistry, 2 ed, John Wiley & Sons, Inc Hoboken, NJ, 2003
Oxidation of antibactenals: supporting information 221
35 Huber, M M, Canonica, S, Park, G-Y, von Gunten, U, Oxidation of
pharmaceuticals during ozonation and advanced oxidation processes
Environ. Sa. Technol. 2003, 37, 1016-1024
36 Qiang, Z M, Adams, C, Potentiometnc determination of acid
dissociation constants (pK(a)) for human and veterinary antibiotics Water
Res. 2004, 38, 2874-2890
37 Hoigné, J, Bader, H, Charactenzation of water-quality critena for
ozonation processes 2 Lifetime of added ozone O^one-Science &
Engineering 199'4, 16, 121-134
38 Leitzke, A , Reisz, E , Flyunt, R, von Sonntag, C ,The reactions of ozone
with cmnamic acids formation and decay of 2-hydroperoxy-2-
hydroxyacetic acid /. Chem. Soc, Perkin Trans. 2 2001, 793-797
39 Lide, D R, Ed CRC Handbook of Chemistry and Physics, 82 ed, CRC Press
Boca Raton, FL, 2001
40 Nash, T, The colonmetric estimation of formaldehyde by means of the
Hantzsch reaction Biochem. J. 1953, 55, 416-421
41 Buxton, G V, Greenstock, W P, Helman, W P, Ross, A B, Cntical
review of rate constants for reactions of hydrated electrons, hydrogen
atoms, and hydroxyl radicals (•OH/,0 ) in aqueous solution J. Phys.Chem. Ref. Data 1988, 17, 513
42 Von Gunten, U, Okveras, Y, Advanced oxidation of bromide-containingwaters Bromate formation mechanisms Environmental Saence & Technology1998, 32, 63-70
43 Stead, D A, Richards, R M E, Sensitive fluonmetnc détection of
gentamicin sulfate in biological matnces using solid-phase extraction, pre-
column denvatization with 9-fluoroenylmethyl chloroformate and
reversed phase high-performance liquid chromatography Journal of
Chromatography B 1996, 675, 295-302
44 Bunton, C A, Oxidation of Ct-diketones and Ct-keto acids by hydrogen
peroxide Nature 1949, 163, 444
45 Josephson, L, Houle, P, Haggerty, M , Stability of dilute aqueous solutions
of gentamicin and tobramycin Clinical Chemistry 1979, 25, 298-300
46 Siegel, B, Lanphear,J, Kinetics and mechanisms for the acid-catalyzed oxidative
decarboxylation of benzylformic acid J. Org. Chem. 1979,44, 942-946
47 Golet, E M, Alder, A C, Giger, W, Environmental exposure and risk
assessment of fluoroquinolone antibacterial agents in wastewater and
river water of the Glatt Valley watershed, Switzerland Environ. Sa. Technol.
2W2,36,3(>45-3(>51
48 Gobel, A, Thomsen, A, McArdeE, C S, Joss, A, Giger, W, Occurrence and
sorption behavior of sulfonamides, macrolides, and trimethoprim in activated
sludge treatment Environmental Science & Technology 2005, 39, 3981-3989
222 Appendix II
49 Elovitz, M S, von Gunten, U, Hydroxyl radical/ozone ratios dunngozonation processes I The Rct concept O^oneSa. Eng. 1999, 21, 239-260
50 Huber, M M, Gobel, A , Joss, A , Hermann, N , Loffler, D, McArdeU, C
S, Ried, A, Siegrist, H, Ternes, T A, von Gunten, U, Oxidation of
pharmaceuticals dunng ozonation of municipal wastewater effluents a
pilot study Environ. Sei. Technol. 2005, 39, 4290-4299
51 Buffle, M -O, Schumacher, J , Salhi, E ,
von Gunten, U, Measurement of
the initial phase of ozone decomposition in water and wastewater bymeans of a continuous quench flow system Application to disinfection
and pharmaceutical oxidation Water Res. 2005, Acceptedforpublication
52 McFarland, J W, Berger, C M , Froshauer, S A, Hayashi, S F, Hecker, S
J, Jaynes, B H, Jefson, M R, Kamicker, B J, Lipinski, C A, Lundy K
M, Reese, C P, Vu, C B, Quantitative structure-activity relationships
among macrolide antibactenal agents In Vitro and in Vivo potency
against Pasteurella multocida J. Med. Chem. 1997, 40, 1340-1346
53 Lucida, H , Parkin, J E, Sunderland, V B, Kinetic study of the reaction
of sulfamethoxazole and glucose under acidic conditions I Effect of pHand temperature Int. J. Pharm. 2000, 202, 47-61
54 Rieder, J, Physikalisch-chenusche und biologische Untersuchungen an
Sulfonamiden 1 Mitteilung Pharmakologisch interessante physikalisch¬chemische Merkmale von 21 Sulfonamiden und 6 Sulfonamid-
Metabokten Ar^neim.-Forsch. 1963, 13, 81-88
55 Vâzquez, J L, Berlanga, M, Merino, S, Domènech, O, Viiïas, M ,
Montero, M T, Hernândez-Borrell, J, Determination by fluonmetnc
titration of the ionization constants of ciprofloxacin in solution and in
the presence of liposomes Photochem. Photobiol. 2001, 73, 14-19
56 Barbosa, J, Barron, D, Jiménez-Lozano, E , Sanz-Nebot, V, Comparisonbetween capillary electrophoresis, liquid chromatography, Potentiometrie,
and spectrophotometnc techniques for evaluation of pKa values of
zwittenonic drugs in acetonitnle-water mixtures Analyiica Chimica Acta
2001,437,309-321
57 Riand, J, Chenon, M T, Lumbrosobader, N, Proton and C-13 Nuclear
Magnetic-Resonance Studies of Substituted Pynmidines 2
Monoprotonation of Methylpynnudines and Ajiunopyrinudines J. Am.
Chem. Soc. 1977, 99, 6838-6845
58 Roth, B, Strelitz, J Z, The Protonation of 2,4-Diaminopyrimidines I
Dissociation Constants and Substituent Effects J. Org. Chem. 1969, 34, 821-836
59 Rapson, H D C, Bird, A E,Ionisation constants of some penicillins
and of their alkaline and penicillinase hydrolysis products J. Pharm.
Pharmacol. 1963, /5, T222-T231
Oxidation of antibactenals: supporting information 223
60 Takâcs-Novâk, K, Box, K J, Avdeef, A, Potentiometnc pi*Cadetermination of water-insoluble compounds validation study in
methanol/water mixtures Int. J. Pharm. 1997, 151, 235-248
61 Stephens, C R, Murai, K, Brunings, K J, Woodward, R B, Acidity constants
of tetracycline antibiotics J.Am. Chem. Soc. 1956, 78, 4155-4158
62 Nieto, M , Perkins, H R, Physicochenucal properties of vancomycia and
lodovancomycin and their complexes with diacetyl-L-lysyl-D-alanyl-D-alanine Biochem. J. 1971, 123, 773-&
63 Antipas, A S, Vandervelde, D, Stella, V J, Factors affecting the deanudation
of vancomycin in aqueous-solutions Int. J. Pharm. 1994, 109, 261-269
64 Kane, R S, Ghnk, P T, Chapman, R G, McDonald, J C, Jensen, P K, Gao,
H, Pasa-Tohc, L, Smith, R D, Whitesides, G M, Basicity of the amino
groups of the aminoglycoside amikacin using capillary electrophoresis and
coupled CE-MS-MS techniques Anal. Chem. 2001, 73, 4028-4036
65 Chnstensen, J J, Izatt, R M , Wrathall, D P, Hansen, L D, Thermodynamicsof proton ionization in dilute aqueous solution Part XI pK, AH°, and AJT°
values for proton ionization from protonated amines at 25° Journal of the
Chemical SocietyA: Inorganic, Physical, Theoretical 1969,1212-1223
66 Weber, E J, Kenneke, J F SPARC
(http //wwwepa gov/athens/research/projects/sparc) US EPA, National
Exposure Research Laboratory Athens, GA
67 Pankratov, A N, Uchaeva, I M, Doronin, S Y, Chernova, R K,
Correlations between the basicity and proton affinity of substituted
anilines /. Struct. Chem. 2001, 42, 739-7'46
68 Hall, H K, Jr, Potentiometnc determination of the base strengths of
amines in non-protolytic solvents J. Phys. Chem. 1956, 60, 63-70
69 Barbosa, J, Barron, D, Cano, J, Jiménez-Lozano, E, Sanz-Nebot, V,
Toro, I, Evaluation of electrophoretic method versus chromatographic,
potentiometnc, and absorptiometnc methodologies for determining pi*Cavalues of quinolones in hydroorganic mixtures J. Pharm. Biomed. Anal.
2001, 24, 1087-1098
70 Hartmann, A , Golet, E M , Gartiser, S, Alder, A C
, Koller, T, Widmer,
R M, Primary DNA damage but not mutagenicity correlates with
ciprofloxacin concentrations in German hospital wastewaters Arch.
Environ. Contam. Toxicol. 1999, 36, 115-119
71 Golet, E M , Xifra, I, Siegnst, H , Alder, A C, Giger, W, Environmental
exposure assessment of fluoroquinolone antibactenal agents from sewage
to soil Environ. Sa. Technol. 2003, 37, 3243-3249
72 Lindberg, R, Jarnheimer, P-A, Olsen, B, Johansson, M, Tyskknd, M,Determination of antibiotic substances in hospital sewage water using
224 Appendix II
solid phase extraction and liquid chromatography/mass spectrometry and
group analogue internal standards Chemosphere 2004, 57, 1479-1488
73 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 Environ. Sei. Technol. 2002, 36, 1202-1211
74 Miao, X-S, Bishay, F, Chen, M, Metcalfe, C D, Occurrence of
antimicrobials in the final effluents of wastewater treatment plants in
Canada Environ. Sa. Technol. 2004, 38, 3542-3550
75 Renew, J E, Huang, C -H, Simultaneous determination of
fluoroquinolone, sulfonamide, and trimethoprim antibiotics in wastewater
using tandem sohd phase extraction and liquid chromatography—
electrospray mass spectrometry/. Chromaiogr, A 2004, 1042, 113-121
76 Ohlsen, K, Ternes, T A, Werner, G, Wallner, U, Loftier, D, Wilma, Z , Witte,
W, Hacker, J, Impact of antibiotics on conjugational resistance gene transfer in
Staphylococcus aureus in sewage Environ. Microbiol. 2003, 5, 711-716
77 Hirsch, R, Ternes, T A, Haberer, K, Kratz, K -L, Occurrence of
antibiotics in the environment Set. Tot. Environ. 1999, 225, 109-118
78 Hartig, C, Storm, T, Jekel, M, Detection and identification of
sulphonamide drugs in municipal waste water by liquid chromatography
coupled with electrospray ionisation tandem mass spectrometry J.
Chromaiogr, A 1999, 854, 163-173
79 Metcalfe, C D, Miao, X-S, Koenig, B G, Struger, J, Distnbution of acidic
and neutral drugs in surface waters near sewage treatment plants in the lower
Great Lakes, Canada Environ. Toxicol. Chem. 2003, 22, 2881-2889
80 Lindberg, R H, Wennberg, P, Johansson, M I, Tyskknd, M , Andersson,
B A V, Screening of human antibiotic substances and determination of
weekly mass flows in five sewage treatment plants in Sweden
Environmental Science & Technology 2005, 39, 3421-3429
81 Andreozzi, R, Caprio, V, Cirugha, C, de Champdoré, M, lo Giudice, R,
Marotta, R, Zuccato, E,Antibiotics in the environment occurrence in
Italian STPs, fate, and preliminary assessment on algal toxicity of
amoxicillin Environ. Sa. Technol. 2004, 38, 6832-6838
82 McArdell, C S, Molnar, E , Suter, M J F, Giger, W, Occurrence and fate of
macrohde antibiotics in wastewater treatment plants and in the Glatt ValleyWatershed, Switzerland Environmental Science e^ Technology 2003, 37, 5479-5486
83 Loffler, D, Ternes, T A, Analytical method for the determination of the
aminoglycoside gentamicin in hospital wastewater via liquid
chromatography—electrospray-tandem mass spectrometry J. Chromaiogr,^4 2003, 1000, 583-588
Marc-Olivier Buffle
Citizen of Canada, and Vandoeuvres, Geneva, Switzerland
1970
1974-1982
1982-1985
1985-1986
1986-1987
1987-1989
1989-1992
1992-1993
1993-1995
1995
1995
1996-98
1998-2001
2001-2005
Date of birth, Geneva
Elementary School, Geneva
Junior High, Geneva
ITigh School, Geneva
ITigh School, Lubbock, Texas, USA
High School, Geneva
Studies in Structural Engineering, ETH Zurich
Junior Engineer, Petignat&Narbel, Montreux, Switzerland
Studies in Structural Engineering, ETH Zurich
M.Eng Thesis, Institute of Building Physics, ETH Zurich
Research at Dipart. di Energetica, Politechnico di Milano, Italy
M.Eng in Structural Engineering, ETH Zurich
Research Engineer, Turbulence Lab, Mech. Eng, U. ofToronto
CED Team Leader, Tro|an Technologies, London, Canada
PhD, Dnnking Water Chemistry, Eawag and ETH Zurich
Photographsoftheblueheronandwaterfountainareadaptedfrom
pictureslicensedundera
CreativeCommons
License
GiardiaLamblia
isadaptedfromaSEMphotographby
ArturoGonzalesCINVESTAVMexicoThephotographofthefrog
wastakenbyReneKüppersGermany
Thephotograph
ofthewatertreatmentplant
isadaptedfroma
pictureof
theHeemskerkWTPand
isproperty
ofPWN
Netherbnd
Photographofthediffuser
isproperty
ofOzonia
Ltd
Picturesoftheozonegeneratorweretakenby
UrsvonGunten