Effect of medium-pressure UV irradiation on bromate concentrations in drinking water, a pilot-scale study

Water Research 38 (2004) 211–217

ARTICLE IN PRESS

*Correspond

fax: +1-519-74

E-mail addr

0043-1354/$ - se

doi:10.1016/j.w

Effect of medium-pressure UV irradiation on bromateconcentrations in drinking water, a pilot-scale study

Sigrid Peldszusa,*, Susan A. Andrewsb, Rosana Souzaa, Franklyn Smithc,Ian Douglasd, Jim Boltone, Peter M. Hucka

aNSERC Chair in Water Treatment, University of Waterloo, Waterloo, Ont. N2L 3G1, CanadabDept. of Civil Engineering, University of Waterloo, Waterloo, Ont. N2L 3G1, Canada

cRegional Municipality of Waterloo, 150 Frederick St. 7th Floor, Kitchener, Ont. N2G 4J3, CanadadCity of Ottawa, Water Division, 2731 Cassels Road, Ottawa, Ont. K2B 1A8, Canada

eBolton Photosciences Inc., 628 Cheriton Cres. NW, Edmonton, Alberta T6R 2M5, Canada

Received 23 October 2002; received in revised form 28 August 2003; accepted 12 September 2003

Abstract

This study investigated the potential for bromate removal from drinking water on irradiation with medium-pressure

UV lamps—a technique gaining considerable interest for drinking water disinfection. Waters from two different sources

were spiked with 20 mg/L of bromate and irradiated with UV fluences up to 718mJ/cm2 utilizing a pilot-scale reactor

(Calgon Carbon Corp.) at a flow of 76L/min (20 gallon/min). Essentially no removal was observed in one of the source

waters. Limited bromate removal, up to 19%, was observed in the second source water at high UV fluences (696mJ/

cm2) and a fluence-response relationship was clearly evident. All removals would be negligible at UV fluences

anticipated for drinking water disinfection (p40mJ/cm2). Different water characteristics, in particular competitive

absorption by nitrate and possibly DOC, were most likely responsible for the differences in bromate removal in the

waters tested. The source water that did not show any removal had a higher nitrate concentration (4 vs. 0.1mg N/L)

and also a higher DOC concentration (4.1 vs. 3.1mg C/L) than the other source water which showed 19% bromate

removal.

r 2003 Elsevier Ltd. All rights reserved.

Keywords: Bromate; Drinking water; Medium-pressure UV–UV disinfection

1. Introduction

Bromate is formed as an ozonation by-product in

bromide-containing waters (e.g. [1,2]). Therefore, water

utilities that utilize ozonation as part of their treatment

process have to be concerned with the possibility of

elevated bromate concentrations. Bromate formation

depends on various factors including initial bromide

concentration, ozone dose, ozone residual and pH.

Bromate concentrations of up to 60mg/L have been

ing author. Tel.: +1-519-888-4567 ext. 3511;

6-7499.

ess: [email protected] (S. Peldszus).

e front matter r 2003 Elsevier Ltd. All rights reserve

atres.2003.09.010

observed following ozonation [3]. As a suspected

carcinogen, bromate is regulated in the United States

to 10 mg/L as of December 2001 [4]. Bull et al. [5]

reported a cancer risk of 10�4 (lifetime exposure) at a

bromate concentration of 5 mg/L, whereas the World

Health Organisation (WHO) [6] has a provisional

guideline value of 25 mg/L based on an excess lifetime

cancer risk of 7� 10�5.

Different approaches have been taken to reduce

bromate concentrations in drinking water. The majority

of studies have focused on optimizing ozonation

processes to minimize bromate formation (e.g., [3,7,8]).

In addition, removal of bromate after its formation has

also been investigated, although to a lesser extent [9–13].

d.

ARTICLE IN PRESSS. Peldszus et al. / Water Research 38 (2004) 211–217212

In the last few years UV disinfection of drinking water

has attracted increased interest, since it has been

demonstrated that Cryptosporidium parvum can be

inactivated by applying UV irradiation with either

low- or medium-pressure lamps at relatively low fluences

[14,15]. Waterworks employing ozone for reasons other

than disinfection (e.g. colour removal, taste and odour

control) may also use UV as their primary disinfection

process. One of the participants in this study is actually

in the process of installing full scale UV downstream of

their ozone application. As a consequence, in bromide-

containing waters, bromate may be present in the

influent of UV reactors.

However, little work has been done on bromate

removal in water by UV irradiation. Siddiqui et al. [9,10]

reported that application of low-pressure UV lamps in

bench scale studies led to lower bromate concentrations.

Mills et al. [11] investigated bromate removal by

semiconductor catalysis in the presence of low-pressure

UV irradiation. In control experiments, when employing

low-pressure UV alone, without the addition of the TiO2

catalyst, they observed bromate removals up to 80%

although only after irradiation times of almost 3 h [11].

Low-pressure UV lamps typically emit light mostly

around 254 nm, whereas medium-pressure UV lamps

emit light to different degrees over a wide range

including lower wavelengths (Fig. 1). Bromate absorbs

UV light well at low wavelengths (Fig. 1) and hence,

photolysis of bromate may be expected when using

medium-pressure UV lamps. Photolysis of bromate

should not be expected when using low-pressure UV

lamps, since bromate does not absorb significantly at

254 nm.

In the work described herein, medium-pressure UV

lamps were used to determine their effect on bromate

concentrations in drinking water. Filter effluents of two

full-scale water treatment plants were spiked with

bromate or a combination of bromate and bromide

and then treated with a pilot-scale UV reactor (Calgon

0

0.1

0.2

0.3

0.4

190 210 230 2Wavelength

Ab

sorb

ance

(A

U)

potassium bromate

medium pressure lamp

low pressurel amp

Fig. 1. UV spectrum of a potassium bromate solution (0.1mM, acqu

low- and medium-pressure UV lamps [20].

Carbon Corp.). Note that reducing the output of

medium-pressure lamps at low wavelength is a legal

requirement in some European countries [16,17]. How-

ever, the medium-pressure lamps in this study utilized

the full wavelength range as is common practice in

North America.

2. Experimental

2.1. Materials and methods

This study utilized a 111L (29.4 gal) Calgon Carbon

Corporation Sentinelt UV reactor containing six 1 kW

medium-pressure UV lamps mounted horizontally

across a vertical reactor (Fig. 2). Lower wavelength

emissions were not blocked out. The reactor was

installed at the Mannheim Water Treatment Plant,

Kitchener (Ontario, Canada) which draws its water

from the Grand River. The plant practices conventional

treatment using coagulation (acidified alum), floccula-

tion and sedimentation followed by ozonation. Feed

water for the reactor was taken following full-scale

biological filters which were immediately preceded by

the ozonation step. For comparison some experiments

employed treated Ottawa River water. The Ottawa

River has different water quality characteristics than

does the Grand River (Table 1). For these experiments,

water from the Britannia Water Purification Plant in

Ottawa (Ontario, Canada) was trucked to Mannheim

where it was fed into the UV reactor that was installed

there. The Britannia water was taken from a clearwell

immediately after filtration. The processes upstream of

filtration are coagulation using alum, flocculation and

sedimentation. Approximately 2mg/L of pre-chlorine is

added in the raw water intake pipe before flocculation.

The general procedure for using the UV unit involved

first adjusting the water flow to the unit, usually 76L/

min (20 gal/min). The UV lamps were switched on and

50 270 290 (nm)

0

3

6

9

12

Lam

p E

mit

tan

ce (

rel)

ired at University of Waterloo) and emission spectra for typical

ARTICLE IN PRESS

Fig. 2. Calgon Carbon Corporation Sentinelt pilot-scale UV reactor.

Table 1

Filtered water characteristics

Mannheim Britannia

Source Grand River Ottawa River

Major impacts Moderately impacted by agriculture,

industry and municipalities

Largely unimpacted with minor logging

activities in the upper river

pHa 7.4 6.5

Alkalinity 150–180mg/L as CaCO3 7–9mg/L as CaCO3

Turbidity 0.04–0.10 NTU o0.05 NTU

TOCa 3.9–4.7mg C/L 3.9mg C/L

DOC 3.6–4.5mg C/L 2.8–3.4mg C/L

UV–absorbancea (254 nm, d ¼ 1 cm) 0.05 AU 0.05 AU

Bromatea oMRLb oMRLb

Bromidea 0.05mg/L 0.01mg/L

Nitratea 3.8–5.4mg N/L 0.1–0.4mg N/L

aData measured at University of Waterloo.

All other data provided by the Region of Waterloo and the City of Ottawa.bMRL: Minimum Reporting Level for bromate 1.1 mg/L.

S. Peldszus et al. / Water Research 38 (2004) 211–217 213

run until they provided a stable output. Then the reactor

influent was spiked with bromate or a combination of

bromide and bromate. A laboratory prepared, concen-

trated solution (bromate 25mg/L, bromide 1262mg/L)

using potassium bromate (ACS grade, BDH) and/or

sodium bromide (ACS grade; Fisher Scientific) was

injected into the influent stream a few meters upstream

of the reactor typically at a flow of 60mL/min. A static

mixer placed about 4m before the reactor intake

ensured adequate mixing of the spiking solution into

the influent stream. The reactor influent target concen-

trations for bromate and bromide were 20mg/L and

1mg/L, respectively. Different UV fluences were

achieved by varying the number of lamps in operation,

generally proceeding from high to low fluences.

For each fluence, the reactor influent and effluent

were sampled in duplicate for bromate and bromide.

Frequent influent sampling allowed for the determina-

tion of the actual influent concentration thus accounting

for variations in the influent concentration due to slight

changes in the flows of the spiking solution and/or the

reactor influent. The measured influent concentrations

were within 86–114% of the target concentrations. In

addition, background samples (no irradiation, no

spiking) were taken for bromate, bromide and total

organic carbon (TOC). Samples were stored in 20 and

40mL glass vials with Teflons-lined caps in a cooler on

icepacks. Upon return to the laboratory, bromide and

bromate samples were immediately analyzed by ion

chromatography (Dionex 300) with a post column

derivatization method [18]. The minimum reporting

levels in our laboratory were 12mg/L for bromide and

1.1 mg/L for bromate. Method accuracy and precision

were determined by measuring seven replicates at two

ARTICLE IN PRESS

-25

0

25

50

75

100

0 100 200 300 400 500 600 700 800

Fluence (mJ/cm2)

Bro

mat

e R

emo

val (

%)

18/04/00, run a

18/04/00, run b

17/05/00, run a

17/05/00, run b

40 mJ/cm2

Fig. 3. Bromate removal upon UV treatment with medium-

pressure lamps in Mannheim filter effluent (run a: spiked with

20 mg/L bromate; run b: spiked with 20mg/L bromate and 1mg/

L bromide; vertical lines=(removalaverage)7[(removal1)�(removalaverage)]).

S. Peldszus et al. / Water Research 38 (2004) 211–217214

different concentration levels. Bromide was measured at

200 and 1500 mg/L and yielded recoveries and standard

deviations of 11276% and 10975%, respectively.

Bromate concentration levels were 7.5 and 30 mg/L with

recoveries and standard deviations of 95710% and

91710%. TOC was analyzed by using combustion

methodology [19]. All parameters including pH were

sampled and measured in duplicate. Quality control

standards were measured in all of our sample runs and

were in the expected concentration range.

The averages for influent and effluent concentrations

were used for the calculation of the removals. However,

duplicate measurements were not sufficient for statistical

treatment of the data and measuring a larger number of

samples per fluence was not feasible. Hence, the

uncertainty of the data was estimated by subtracting

the difference of data set 1 from the difference of the

averages [i.e., error=(effl.average–infl.average)7{(effl.1�infl.1)�(effl.average�infl.average)}]. For each fluence, this

value is depicted as a vertical line in the figures.

2.2. UV fluence calculations

In this research, the term fluence is used rather than

the term UV dose. Fluence is expressed in mJ/cm2 (other

authors have used the equivalent mWs/cm2 unit). The

fluence (mJ/cm2) was obtained by multiplying the

average fluence rate (mW/cm2) in the water by the

irradiation time (min) [20]. These fluences represent an

ideal or maximum absorbed fluence, which were likely

too high due to imperfect mixing in the reactor. Hence,

the actual fluences of the reactor were determined by

biodosimetry using MS2 coliphage [21,22]. In order to

relate the calculated maximum fluence to the actual

fluence determined by biodosimetry, a correction factor

of 0.2 was applied to all of the calculated maximum

fluences. The resulting corrected fluence was used

throughout this paper. These corrections were necessary,

since the relatively low flow rate of 76L/min (20 gal/min)

leads to incomplete mixing in the reactor.

The fluences employed ranged from 0 to 718mJ/cm2.

Although fluences required for drinking water disinfec-

tion could be equal or less than 40mJ/cm2, higher

fluences were included as part of this study to see if any

trends could be observed.

2.3. Selection of spiking conditions

Both source waters had very low background bromide

concentrations, and bromate was not detected at all

(Table 1). Bromate was spiked into the reactor influent

at a concentration of 20 mg/L. This concentration was

chosen to represent a compromise between the regulated

value of 10mg/L in the United States [4] and the WHO

provisional guideline value of 25 mg/L [6]. The spiking of

bromate alone allowed not only the monitoring of

potential bromate removal, but also for measuring

bromide production which is reported to be the end

product of UV induced bromate degradation [9,10]. In

other experiments, bromide (1mg/L) was spiked to-

gether with bromate (20 mg/L). As an ozonation by-

product, bromate is formed from bromide, which is

typically present in higher concentrations than bromate.

Hence, experiments with simultaneous bromide and

bromate additions were more representative of ozo-

nated, bromide-containing waters than experiments with

bromate spiking alone.

3. Results

For each fluence bromate and bromide concentrations

were measured in duplicate in both the reactor effluent

and influent. The averages for influent and effluent

concentrations were used for the calculation of the

removals.

In general, the UV treatment of spiked Mannheim

filter effluent did not show any substantial reductions in

bromate concentration (Fig. 3), even at high fluences.

Removals of bromate fluctuated between –5.4% and

+21.1% with no apparent fluence relationship. Runs

with bromate spiking alone produced similar results to

those with combined bromate and bromide spiking.

Fig. 4 shows trends for bromate and bromide for the

experiment where bromate was spiked alone (20 mg/L=0.156 mM). Molar concentrations are given since

photolysis of 1mol of bromate results in the production

of 1mol of bromide [9] and ideally, any removal of

bromate should be accompanied by the corresponding

formation of bromide. However, Mannheim water had

an average background concentration of 47 mg/L of

bromide, whereas bromate was not detected. This

ARTICLE IN PRESS

-0.15

-0.10

-0.05

0.00

0.05

0.10

0 100 200 300 400 500 600 700 800

Fluence (mJ/cm2)

Eff

luen

t -

Infl

uen

t (µ

M)

bromatebromide

40 mJ/cm2

Fig. 4. Behavior of bromate and bromide after UV treatment

with medium-pressure lamps in Mannheim filter effluent water

(spiked with 20mg/L=0.156 mM bromate; vertical lines=

(effl.average–infl.average)7[(effl.1�infl.1)�(effl.average–infl.average)]).

y = -5E-05x + 0.0046

R2 = 0.7021

-0.15

-0.10

-0.05

0.00

0.05

0.10

0 100 200 300 400 500 600 700 800

Fluence (mJ/cm2)

Eff

luen

t -

Infl

uen

t (µ

M)

bromate, run a

bromide, run a

bromate, run b

40 mJ/cm2

Fig. 5. Behavior of bromate and bromide after UV treat-

ment with medium-pressure lamps in Britannia water

(run a: spiked with 20 mg/L=0.156mM bromate; run b:

spiked with 20 mg/L=0.156mM bromate and 1mg/L=13 mMbromide; vertical lines=(effl.average–infl.average)7[(effl.1�infl.1)�(effl.average–infl.average)]).

S. Peldszus et al. / Water Research 38 (2004) 211–217 215

bromide background made it difficult to observe any

bromide formation which would be in the range of only

a few mg/L. Nevertheless, for most data points a slight

reduction in bromate is accompanied by the formation

of bromide. The bromide data point at a fluence of

685mJ/cm2 was exceptionally high—possibly due to

some short-term fluctuations in the flow of the spiking

pump. For this particular data point the bromide

influent concentration was 64 mg/L, while the average

influent concentration for all the other data points was

47mg/L. Incomplete mass balances may also be caused

by the formation of intermediate species which were not

accounted for in these experiments.

Fig. 5 shows the results for the Britannia water. When

high UV fluences were applied, bromate concentrations

decreased moderately in both the presence and absence

of bromide. When bromate was spiked alone (20 mg/L=0.156 mM), a corresponding increase in bromide

concentration could be observed for most data points

despite a bromide background concentration in the

influent of 11 mg/L. Again, bromate was not detected in

the influent. In this source water, up to 19% of bromate

was removed at a fluence of 696mJ/cm2. At lower

fluences, removals were in the same range as observed

for Mannheim water. The Britannia water exhibited a

clear fluence response relationship, which was not the

case for the Mannheim water. However, at fluences

applicable for drinking water disinfection, that is,

p40mJ/cm2, bromate removals in Britannia water

should be negligible.

The different behavior of bromate in response to

medium-pressure UV irradiation in the two source

waters may be attributed to their different water quality

characteristics. When undergoing direct photolysis,

bromate absorbs the UV light directly and undergoes

radical reactions ultimately leading to the reduction of

bromate to bromide [9]. In a matrix such as drinking

water, bromate has to compete with other molecules for

the UV light, thus lowering the effects of direct

photolysis of bromate in the presence of these com-

pounds. Upon UV absorption competing molecules

such as nitrate or organic material can form radical

oxidizing species which in turn can reoxidize bromide or

any other bromine containing intermediates back to

bromate. This process is known as indirect photolysis

[23]. Hence, overall bromate removal by UV irradiation

is the net result of the reduction of bromate to bromide

by direct photolysis and the reoxidation of bromide and

intermediate bromine species by radical oxidative species

which are formed through photolysis of other competing

compounds. However, only little is known about these

reactions, which makes mechanistic data interpretation

difficult.

The different characteristics of Mannheim vs. Brit-

annia water are summarized in Table 1. The Britannia

water had a lower pH of 6.5 and a much lower alkalinity

(9mg/L as CaCO3) than the Mannheim water (pH 7.4;

alkalinity 150mg/L as CaCO3). The TOC values were

comparable (3.9mg C/L), whereas the DOC was lower

in the Britannia water (2.8–3.4mg C/L) than in the

Mannheim water (3.6–4.5mg C/L). Nitrate concentra-

tions differed substantially with 3.8–5.4mg N/L for

Mannheim and 0.1–0.4mg N/L for Britannia water.

Competitive UV absorption by nitrate may have

contributed to the fact that no bromate removal was

observed in Mannheim water whereas some was seen in

Britannia water. Direct photolysis of nitrate leads to the

formation of nitrite [24] and is typically only observed

upon irradiation with medium-pressure UV lamps,

which do not have their lower wavelength blocked out.

In all experiments with both waters nitrite was formed

upon UV irradiation. Nitrite concentrations were a

factor of 10 higher in Mannheim water (197mg N/L vs.

ARTICLE IN PRESSS. Peldszus et al. / Water Research 38 (2004) 211–217216

20mg N/L for Britannia), most likely because of its

higher nitrate concentration. These maximum nitrite

concentrations were only observed at high fluences and

were well below the regulated limit. Both source waters

displayed a clear fluence response relationship for the

formation of nitrite (data not shown).

Furthermore, the lower dissolved organic carbon

(DOC) concentrations in the Britannia water may have

also contributed to the higher removals observed in this

source water. This is consistent with previous reports

which stated that ‘the presence of DOC lowered the

bromate reduction and decreased the production of

bromide in natural waters during UV irradiation’ [9].

Competitive absorption of UV light by the organic

material over bromate may be an explanation for this

phenomenon. As a consequence, simultaneous forma-

tion of organic radicals can take place with subsequent

formation of e.g. �OH, 1O2 or H2O2 [23]. These are all

oxidative radical species which may reoxidize bromide

or any other bromine containing intermediates back to

bromate. To further complicate the situation, carbonate

and other inorganic anions can act as scavengers for the

various radicals namely OH radicals. At this point, the

higher alkalinity in the Mannheim water may have had

an effect on the overall reactions. In general, due to the

complexity of the radical chain reactions overall net

results of bromate photolysis cannot be predicted by

water quality parameters alone, thus emphasizing the

importance of bench and pilot testing of water treatment

processes.

4. Conclusions

This pilot-scale study investigated the potential

bromate removal when irradiating two different drink-

ing waters with medium-pressure UV light. The results

may be summarized as follows.

* Bromate was not removed in one water (Mannheim

filter effluent), although fluences up to 718mJ/cm2

were applied.* Bromate removals up to 19% were observed in the

second treated water (Britannia), although only at

high UV fluences (>300 to 696mJ/cm2). At fluences

typical for drinking water disinfection (p40mJ/cm2)

this effect would be negligible.* The main water quality characteristics that differed

between the test waters were nitrate level, alkalinity,

and to a lesser degree DOC. Competitive absorption

by nitrate and possibly DOC may have lead to a less

effective reduction in bromate concentration by

photolysis. Overall net results of bromate photolysis

cannot be predicted from water quality parameters

alone because of the complexity of the radical chain

reactions.

* There were no differences in effluent bromate

concentrations between runs in which bromate was

spiked by itself and runs with simultaneous bromate

and bromide spiking.

Acknowledgements

Funding for this project was provided by the Natural

Sciences and Engineering Research Council of Canada

(NSERC) in the form of a Collaborative Research and

Development (CRD) grant. The industrial partner for

this project was Calgon Carbon Corporation, which

provided the UV reactor and also financial support.

Some funding was received through the NSERC

Industrial Research Chair in Water Treatment at the

University of Waterloo. The Chair partners included:

American Water Services Canada Corp., the City of

Brantford, the City of Ottawa, the City of Toronto,

Conestoga-Rovers & Associates Limited, EPCOR

Water Services, NSERC, the Ontario Clean Water

Agency, PICA USA Inc., RAL Engineering Ltd., the

Regional Municipality of Waterloo, Stantec Consulting

Ltd., the University of Waterloo and the Windsor

Utilities Commission.

References

[1] Haag WR, Hoign!e J. Ozonation of bromide containing

waters: kinetics of formation of hypobromous acid and

bromate. Environ Sci Technol 1983;17:261–7.

[2] Von Gunten U, Hoign!e J. Bromate formation during

ozonation of bromide-containing waters: interaction of

ozone and hydroxyl radical reactions. Environ Sci Technol

1994;28:1234–42.

[3] Krasner SW, Glaze WH, Weinberg HS, Daniel PA, Najm

IN. Formation and control of bromate during ozonation

of waters containing bromide. J Am Water Works Assoc

1993;85(1):73–81.

[4] United States—Environmental Protection Agency: Na-

tional primary drinking water regulations: disinfectants

and disinfection byproducts. Federal Register 1998;63(241)

Dec. 16, 69390–476.

[5] Bull RJ, Krasner SW, Daniel, PA Bull RD. Health effects

and occurrence of disinfection by-products. American

Waterworks Association Research Foundation Report,

Denver, CO, USA, 2001.

[6] World Health Organization. Guidelines for drinking water

quality, vol. 2, Health criteria, 2nd ed., 1996; Geneva,

www.who.int/water sanitation health/GDWQ (Oct. 2003).

[7] Song R, Westerhoff P, Minear R, Amy G. Bromate

minimization during ozonation. J Am Water Works Assoc

1997;89(6):69–78.

[8] Hofmann R, Andrews RC. Ammonical bromamines: a

review of their influence on bromate formation during

ozonation. Wat Res 2001;35(3):599–604.

ARTICLE IN PRESSS. Peldszus et al. / Water Research 38 (2004) 211–217 217

[9] Siddiqui M, Amy G, Ozekin K, Zhai W, Westerhoff P.

Alternative strategies for removing bromate. J Am Water

Works Assoc 1994;86(10):81–96.

[10] Siddiqui M, Amy G, Zhai W, Ozekin K, Westerhoff P,

Chandrakanth M. Removal of bromate after ozonation

during drinking water treatment. Proceedings American

Water Works Association Annual Conference, New York,

NY, 1994. p. 881–905.

[11] Mills A, Belghazi A, Rodman D. Bromate removal from

drinking water by semiconductor photocatalysis. Wat Res

1996;30(9):1973–8.

[12] Hijnen WAM, Jong R, van der Kooij D. Bromate removal

in a denitrifying bioreactor used in water treatment. Wat

Res 1999;33(4):1049–53.

[13] Gordon G, Gauw RD, Emmert GL, Walters BD, Bubnis

B. Chemical reduction methods for bromate ion removal. J

Am Water Works Assoc 2002;94(2):91–8.

[14] Bolton JR, Dussert B, Bukhari Z, Hargy T, Clancy JL.

Inactivation of Cryptosporidium parvum by medium-

pressure ultraviolet light in finished drinking water.

Proceedings, American Water Works Association Annual

Conference, Dallas, TX, vol. A, 1998. p. 389–403.

[15] Clancy JL, Bukhari Z, Hargy TM, Bolton JR, Dussert

BW, Marshall MM. Using UV to inactivate Cryptospor-

idium. J Am Water Works Assoc 2000;92(9):97–104.

[16] DVGW. German association on gas and water technical

standard W 294: UV systems for disinfection in drinking

water supplies—requirements and testing. DVGW stan-

dards and Guidelines, Bonn, Germany, 1997, http://

www.DVGW.de.

[17] .ONORM. Plants for the disinfection of water using

ultraviolet radiation: Requirements and testing..Osterreichisches Normungsinstitut, Vienna, Austria,

2001, http://www.on-norm.at.

[18] Salhi E, von Gunten U. Simultaneous determination of

bromide, bromate and nitrite in low mg/L levels by ion

chromatography without sample pretreatment. Wat Res

1999;33(15):3239–44.

[19] American Public Health Association, American Water

Works Association, Water Environment Federation.

Standard methods for the examination of water and

wastewater, 20th ed., 1998.

[20] Bolton JR, Linden KG. Standardization of methods

for fluence (UV dose) determination in bench-

scale UV experiments. J Environ Engr 2003;129(3):

209–16.

[21] Bolton JR. Report on UV dose test carried out on the

Calgon Carbon UV Reactor at the Mannheim Water

Treatment Plant. Report for Calgon Carbon Corporation,

Bolton Photosciences Inc., 628 Cheriton Cres. NW,

Edmonton, Alberta, Canada T6R 2M5, 2001.

[22] NWRI/AWWARF. Ultraviolet Disinfection Guidelines

for Drinking Water and Water Reuse. National

Water Research Institute, Fountain Valley, CA, USA,

2000.

[23] Halmann MM. Photodegradation of water pollutants.

Boca Raton, FL, USA: CRC Press; 1996.

[24] Mack J, Bolton JR. Photochemistry of nitrite and nitrate

in aqueous solution: a review. J Photobiol Photochem A

1999;128:1–13.