INACTIVATION OF ESCHERICHIA COLI WITH OZONE:

CHEMICAL AND INACTIVATION KINETICS

NIMRATA K. HUNT

1

and BENITO J. MARINÄAS

2

*

*

M

1

Woodard and Curran Environmental Services, Portland, ME 04102, USA;

2

Department of Civil and

Environmental Engineering, University of Illinois, Urbana, IL 61801, USA

(First received August 1998; accepted in revised form December 1998)

AbstractÐThe apparent chemical and inactivation reactions taking place during the disinfection of

Escherichia coli with ozone in the presence of humic acid were investigated with continuous-¯ow tubu-

lar reactors. The apparent decomposition of dissolved ozone in the presence of humic acid and E. coli

cells was modeled successfully with mixed second-order rate expressions within a time scale relevant to

E. coli inactivation with ozone. The rate for the ozone inactivation of E. coli in the presence of humic

acid was slower than that in the absence of natural organic matter due to the faster decomposition of

dissolved ozone and thus the lower exposure of E. coli cells to the disinfectant in the former case. How-

ever, the second-order inactivation rate constant was approximately the same in the presence and

absence of humic acid con®rming that molecular ozone rather than radicals was the species generally

responsible for inactivation. The overall rate of reaction between ozone and organic matter associated

with E. coli cells was slower that the rate of E. coli inactivation by ozone. Only 25% of the initial rela-

tively fast ozone demand were satis®ed by the time that practically all micro-organisms present in sol-

ution were inactivated. TEM micrographs revealed that noticeable changes in the interior of E. coli

cells did not take place until most of the cells present in the sample were non-viable. # 1999 Elsevier

Science Ltd. All rights reserved

Key wordsÐcontinuous-¯ow reactor, Escherichia coli, humic acid, inactivation kinetics, ozone disinfec-

tion, transmission electron microscopy

INTRODUCTION

The disinfection of Escherichia coli (E. coli) with

ozone has been investigated with several con®gur-

ations of semi-batch and continuous-¯ow through

reactors (Perrich et al., 1975; Zhou and Smith,

1994; Hunt and MarinÄas, 1997). In general, the in-

activation kinetics were found to be pH indepen-

dent and consistent with the classic Chick±Watson

model (Chick, 1908; Watson, 1908) according to the

batch reactor expression:

dN

dt

ˆ ÿk

i

cN,

…1†

where N is the density of viable micro-organisms, c

is the concentration of dissolved molecular ozone,

both at time t from the start of the reaction, and k

i

is the inactivation rate constant given by (Hunt and

MarinÄas, 1997):

k

i

ˆ A exp



ÿ

E

a

RT



…2†

in which E

a

=37,100 J/mol is the activation energy,

A=5.3710

8

l/(mg s) is the frequency factor,

R=8.314 J/(mol K) is the ideal gas constant and T

is absolute temperature in K. Deviations from

equation 1 including inactivation curve shoulders

and tails have been attributed to departure of ex-

perimental units from ideal reactor behavior and

the occurrence of small fractions of bacterial clumps

(Hunt and MarinÄas, 1997).

The consistency of experimental data with

equation 1 indicates that the inactivation kinetics

could be modeled with a homogeneous second

order rate law. However, the inactivation process is

a more complex heterogeneous phenomenon includ-

ing various mass transfer steps and reactions. The

disinfectant must di€use toward the surface of the

micro-organism, and then permeate into the mem-

brane and cytoplasm. The rate of mass transfer

could be a€ected by various intracellular and extra-

cellular reactions with biomolecules. Inactivation or

loss of viability would occur when vital constituents

would su€er a certain level of irreversible damage.

An important concept is that the chemical reactions

between ozone and biomolecules continue for a

relatively longer period of time after loss of viability

until the disinfectant is depleted or the oxidation

demand by biomolecules is satis®ed.

Wat. Res. Vol. 33, No. 11, pp. 2633±2641, 1999

# 1999 Elsevier Science Ltd All rights reserved

Printed in Great Britain

0043-1354/99/$ - see front matter

PII: S0043-1354(99)00115-3

*Author to whom correspondence should be addressed.

[Tel.: [phone]; fax: [phone]; e-mail:

[e-mail]].

2633

Despite decades of research e€orts by various

groups, the mechanisms responsible for the inacti-

vation of E. coli are still not well understood.

Christensen and Giese (1954) suggested that the

bacterial membrane was the ®rst site of ozone

attack. In agreement with this suggestion, Scott and

Lesher (1963) indicated that ozone attacked the cell

surface, altering the permeability of the membrane

and ultimately resulting in leakage of cell contents

into the medium. The occurrence of leakage was

con®rmed by showing that the concentrations of

nucleic acids and proteins increased in the medium,

and decreased inside cells. Though ozone a€ected

the aggregation of nucleoproteins inside the cells, it

did not react appreciably with the nucleic acids

prior to being released into the medium. Further

evidence of the inability of ozone to penetrate the

cells was given by the fact that the highly reactive

reduced sulfur groups of glutathione inside the cells

were una€ected, while in contrast ozone reacted

very fast with glutathione present in the medium.

More recently, other authors have reported that

proteins and unsaturated lipids in the cell mem-

brane are the targets of ozone attack (Pryor et al.,

1983) and leakage of cell contents into the medium

and ultimately cell lysis can result after prolonged

ozonation (Scott, 1975; Hamelin et al., 1978).

Hamelin et al. (1978) suggested that ozone pro-

duced single strand breaks in DNA which, if unre-

paired, caused extensive breakdown of DNA in E.

coli ultimately resulting in the loss of viability.

Besides these lesions, ozone might also induce base

damage and protein cross-linking in DNA mol-

ecules (Hamelin et al., 1977).

Perrich et al. (1975) concluded that cell lysis was

not the main mechanism for inactivation of E. coli

and that the cells remained morphologically intact

after inactivation. Ishizaki et al. (1987) determined

that ozone a€ected plasmid DNA harbored in E.

coli cells by converting the closed circular DNA to

open circular DNA. This observation indicated that

ozone was able to di€use through the cell mem-

brane and react with cell constituents. The authors

proposed that damage to chromosomal DNA might

be one of the reasons for inactivation of E. coli by

ozone. Hamelin and Chung (1974) studied the abil-

ity of ozone to induce mutations in E. coli and pos-

tulated that ozone was able to penetrate into the

cell and genetically alter the cytoplasmic constitu-

ents before destroying the cell membrane. They

determined that inactivation was a function of the

amount of ozone available per bacterium, and

below a certain threshold concentration, ozone had

no e€ect on the survival of bacteria.

Disinfection of E. coli in natural water and

wastewater presents an additional degree of com-

plexity because ozone will also react with dissolved,

colloidal and particulate matter, and these reactions

might interfere with some of the reactions respon-

sible for E. coli inactivation. Designing disinfection

reactors thus might require the simultaneous con-

sideration of all reactions a€ecting the concen-

tration of dissolved ozone and ultimately the

inactivation process.

The main objective of this study was to investi-

gate and model the apparent chemical and inacti-

vation reactions taking place during the disinfection

of E. coli with ozone in the presence of humic acid,

a group of compounds selected to represent a frac-

tion of the organic matter present in natural waters.

An additional objective was the assessment of E.

coli cell structure transformation at various levels of

exposure to ozone.

MATERIALS AND METHODS

Experimental design

The scope of work of this study included the perform-

ance of four experimental sets. Two sets were designed to

investigate separately the apparent overall decomposition

of ozone reacting with E. coli and humic acid. The third

experimental set was performed to con®rm that the appar-

ent kinetics determined for the separate reactions with the

previous two sets could be used to represent the overall

decomposition of ozone in the simultaneous presence of E.

coli and humic acid. The objective of the fourth set was to

determine the e€ect of the presence of humic acid on the

inactivation kinetics of E. coli with ozone. All experiments

were performed with a continuous-¯ow tubular reactor ap-

paratus described previously (Hunt and MarinÄas, 1997). A

summary of experimental conditions used is presented in

Table 1. Experimental set I also included tests performed

with stock solutions of E. coli sonicated in a Fisher 50

Sonic Dismemberator (Fisher Scienti®c, Itasca, IL) to

assess potential mass transfer limitations a€ecting the

overall inactivation process. Complete cell lysis resulting

from the sonication treatment was con®rmed microscopi-

cally.

Table 1. Summary of initial ozone (c

o

), E. coli (N

o

) and humic acid (TOC

o

) concentrations.

Additional conditions applicable to all experimental sets unless otherwise indicated:

temperature=208C,

pH=6,

total

phosphate

concentration=0.02 M,

tert-butanol

concentration=0.01 M, tubular reactor ¯ow rate=50 ml/min, contact time (t)=3.8±48.4 s

Experimental set

c

o

(mg/l)

N

o

(10

9

CFU/l)

TOC

o

(mg/l)

I

65±266

0.35±7.2

0

II

26±304

0

0±1.0

III$

190±374

7.0±13.

0±0.50

IV

6±52

5.7±13

0±0.50

$

Tubular reactor ¯ow rate=65 ml/min, t=2.5±47.8 s.

Nimrata K. Hunt and Benito J. MarinÄas

2634

Materials

Stock solutions of ozone, E. coli and humic acids were

prepared in ozone demand free water bu€ered at pH 6.0

or 7.2 with 0.02 M phosphate. E. coli ATCC strain 11775

(American Type Culture Collection, Rockville, MD) was

used as the test micro-organism. Humic acid was obtained

from

Aldrich

Chemical

Co.

(Milwaukee,

WI).

Experimental solutions were dosed with 0.01 M tert-buta-

nol for radical scavenging and the temperature of all tests

was maintained constant at 208C by means of immersing

the tubular reactors in a water bath. Additional infor-

mation about experimental, chemical and microbial com-

ponents was presented in a previous publication (Hunt

and MarinÄas, 1997).

Analytical methods

Total organic carbon (TOC) analyses were performed to

determine the organic carbon concentrations correspond-

ing to the various humic acid dosages used as well as to

quantify the organic carbon content of E. coli cells. The

latter analyses were performed by sonicating 10 ml of

stock E. coli solution in a chilled water bath for approxi-

mately 10 min, the time required for complete cell lysis as

con®rmed by microscopic examination. A few drops of

concentrated phosphoric acid (85%) were added to the

sonicated cells prior to sample analyses with a TOC analy-

zer Dohrmann Model DC-80 (Tekmar-Dohrmann,

Cincinnati, OH). Analytical procedures for ozone determi-

nation and E. coli viability assessment were described pre-

viously (Hunt and MarinÄas, 1997).

Transmission electron microscopy

Additional inactivation experiments were performed to

study structural changes of E. coli cells after several levels

of exposure to ozone. Experimental conditions included

initial ozone concentrations c

o

=0±1000 mg/l, E. coli den-

sity N

o

=510

9

CFU/l and contact time t=30 s. The ex-

periments were performed at 208C and pH 7.2 in 0.02 M

phosphate bu€er solution in the presence of 0.01 M tert-

butanol. For each test, 50-ml sample volumes were col-

lected in bottles containing 10 ml of 0.2% sodium thiosul-

fate solution to quench excess ozone. The samples were

centrifuged to concentrate the micro-organisms, which

were then ®xed with an equal volume of 3% glutaralde-

hyde (Electron Microscopy Sciences, Fort Washington,

PA) bu€ered at pH 7.2 with phosphate. After approxi-

mately 1 h, the samples were centrifuged and the resulting

bacterial pellets were exposed overnight to additional

phosphate-bu€ered 3% glutaraldehyde solution.

The glutaraldehyde-®xed bacteria were embedded in 1%

agar and washed twice in phosphate bu€er. The samples

were then ®xed in bu€ered 1% osmium tetroxide/1.5%

potassium ferrocyanide at 48C for 1.5 h. The ®xed samples

were rinsed in phosphate bu€er twice and dried by ex-

posure to gradually increasing concentrations of ethanol

(30, 60, 90, 100 and 100%). The dried samples were rinsed

twice in propylene oxide and then in®ltrated with propy-

lene oxide±epoxy resin Poly/Bed 812 (Polysciences,

Warrington, VA) mixtures until the samples were in pure

epoxy resin. The samples were placed in polyethylene cap-

sules and the resin polymerized at 608C for 48 h. Thin sec-

tions were cut on a Reichert-Jung Ultracut E

ultramicrotome (Carl Zeiss, Thornwood, NY) and stained

with uranyl acetate and lead citrate. The sections were

examined with a JEOL JEM-100CX transmission electron

microscope (TEM) (JEOL USA, Peabody, MA) operating

at 80 kV and micrographs were recorded on Kodak SO-

163 electron image ®lm (Eastman Kodak Co., Rochester,

NY) and subsequently digitized with a high-resolution

scanner.

RESULTS AND DISCUSSION

Experimental results obtained for the tests per-

formed to assess the overall decomposition kinetics

of ozone in the presence of E. coli are presented in

Fig. 1. Modeling e€orts revealed that the apparent

ozone decomposition could be represented with a

mixed second order rate expression or

dc

dt

ˆ ÿk

x

cx,

…3†

where k

x

is a second order rate constant in l/(mg s)

and x is the concentration of fast ozone demand

constituents in E. coli cells in mg/l as O

3

given by:

x ˆ c ÿ c

o

‡ x

o

…4†

in which x

o

is the initial concentration of fast

ozone-demand constituents, a quantity proportional

to the initial number of viable micro-organisms N

o

or x

o

=a

o

N

o

. a

o

can be expressed in mg of O

3

per

colony forming unit (mg O

3

/CFU) or the ozone

demand per E. coli cell assuming that each bacterial

colony resulted from an individual cell.

Integration of equation 3 after substitution of

equation 4 resulted in the following expression:

Fig. 1. Apparent overall decomposition of ozone in the

presence of initially viable E. coli for various initial ozone

concentrations and micro-organism densities (see set I in

Table 1 for experimental conditions).

Inactivation of E. coli with ozone

2635

c ˆ

c

o

…c

o

ÿ a

o

N

o

†

c

o

ÿ a

o

N

o

exp… ÿ k

x

…c

o

ÿ a

o

N

o

†t†

…5†

Equation 5 was used to ®t the entire data set pre-

sented in Fig. 1 by a least-square method. The

results from these e€orts are presented in the ®gure

by the continuous lines. The corresponding ®tting

parameter values obtained were k

x

=1.7220.15 l/

(mg s) and a

o

=1.2920.0510

ÿ11

mg O

3

/CFU. As

depicted in Fig. 1, the second-order kinetic model

®tted the experimental data well for t<15 s.

Although deviations were observed for t>15 s,

these longer contact times were not relevant because

the E. coli inactivation reaction is much faster (k

i

/

k

x

180) and practically completed under 15 s for the

entire range of experimental conditions investigated.

TOC analyses of sonicated E. coli cell suspensions

at initial densities ranging from 6.8±1110

9

CFU/l

revealed that the average organic carbon content of

each

cell

was

approximately

TOC

o

/

N

o

=2.5620.1010

ÿ10

mg C/CFU. The ozone

demand per mass of organic carbon can be

expressed with the new parameter b

o

=a

o

N

o

/

TOC

o

=0.050420.0040 mg O

3

/mg C. The ozone

demand parameters could also be expressed on a

molecular base as a

o

=1.6210

8

O

3

molecules/CFU

and b

o

=0.0126 O

3

molecules/E. coli C atom.

Interestingly, the a

o

value for ozone was in reason-

ably good agreement with that of 710

7

molecules/

CFU for iodine demand by E. coli reported by

Scott and Lesher (1963). These authors proposed

that the oxidizing agent demand was exerted by car-

bon double bonds of unsaturated fatty acids present

in the cell wall and membrane. Assuming that each

ozone molecule reacts with a single carbon double

bond group, then the fraction of E. coli organic car-

bon participating in double bond groups would be

approximately 2.52% of the total carbon content.

It was assumed in equation 3 that the rate of

ozone disappearance in the presence of E. coli cells

can be represented by a homogeneous second-order

expression. The validity of this assumption was

checked by testing solutions containing both in-

itially viable, and sonicated E. coli cells. The results

from these experiments are presented in Fig. 2. A

somewhat faster ozone decay rate was observed for

the sonicated samples. As expected, predictions with

equation

3

using

k

x

=1.72 l/(mg s),

and

a

o

=1.2910

ÿ11

mg/CFU (continuous lines) were

closer to the data corresponding to the initially

viable cells. The slight over-predictions observed

might have resulted from a combination of analyti-

cal and experimental inaccuracies. In order to com-

pare these results, equation 3 was used to ®t the

data sets corresponding to the initially live and

sonicated cells (dashed lines). The resulting par-

ameters

were

k

x

=1.3320.14 l/(mg s)

and

a

o

=1.6620.0710

ÿ11

mg/CFU for initially live

cells

and

k

x

=2.3820.36 l/(mg s)

and

a

o

=1.7720.0810

ÿ11

mg/CFU for sonicated cells.

The fact that the a

o

values were comparable sup-

ported that the ultimate ozone demand was ap-

proximately the same for both types of cell

suspensions with the small di€erence obtained likely

being the result of experimental variability. The

di€erent k

x

values were indicative that the inacti-

vation process might have been mass transfer lim-

ited in agreement with disinfection mechanisms

based on reactivity with cell membrane components

(Christensen and Giese, 1954; Scott and Lesher,

1963). Accordingly, the faster reaction rate for soni-

cated cells indicated that ozone molecules reacted

faster with cell components. Consequently, the use

of equation 5 should be interpreted as a model

valid for quanti®cation but not necessarily an accu-

rate representation of the more complex mass trans-

fer steps and heterogeneous reactions taking place

during the inactivation process.

Experimental results obtained for the tests per-

formed to investigate the decomposition of ozone in

the presence of humic acid are presented in Fig. 3

for t<15 s. Equations 3±5 were used to ®t the

entire data set shown in the ®gure after substituting

Fig. 2. Apparent overall decomposition of ozone in the

presence of initially viable and sonicated E. coli for var-

ious initial ozone concentrations and micro-organism den-

sities (see set I in Table 1 for experimental conditions).

Nimrata K. Hunt and Benito J. MarinÄas

2636

x for y, and x

o

=a

o

N

o

for y

o

=b

o

TOC

o

. y is the fast

reacting portion of the TOC also expressed in mg/l

as O

3

, y

o

is the initial value for y, TOC

o

is the TOC

corresponding to the initial humic acid concen-

tration in solution, and b

o

is the initial fast ozone

demand by the humic acid tested also expressed in

mg O

3

/mg C. The ®tted curves were found to

match the experimental data reasonably well as

depicted in Fig. 3. The corresponding ®tting par-

ameters

were

k

y

=3.2020.16 l/(mg s)

and

b

o

=0.32420.005 mg O

3

/mg C. It should be

pointed out that similar to the observation for the

reaction with E. coli, the model deviated from the

experimental data for t>15 s possibly due to ozone

reactions with slower-reacting organic molecules

and radicals. However, as discussed earlier these

longer contact times were generally not relevant to

the inactivation process under the set of experimen-

tal conditions used.

The decomposition of dissolved ozone in the sim-

ultaneous presence of both E. coli and humic acid

can be represented by the expressions:

dc

dt

ˆ ÿk

x

cx ÿ k

y

cy

…6†

dx

dt

ˆ ÿk

x

cx

…7†

dy

dt

ˆ ÿk

y

cy

…8†

Equations 6±8 represent a system of independent

non-linear equations meeting the overall mass bal-

ance (c

o

ÿx

o

ÿy

o

)=(cÿxÿy) which can be used to

substitute any of the three expressions. The non-lin-

earity of the system of equations required solution

by an approximation approach. A ®nite di€erence

approach was used with time interval Dt=0.003 s

or a total of 500 time elements within the range of

0±15 s. A trial-and-error approach revealed that

predictions were practically the same for lower Dt

values. Figure 4 presents a comparison of model

predictions to experimental data corresponding to

tests performed at N

o

=1.310

10

CFU/l, TOC

o

=0±

0.50 mg/l and c

o

=206±374 mg/l. The model par-

ameters were those independently determined by ®t-

ting the data in Figs. 1 and 3 (i.e. k

x

=1.72 l/(mg s),

k

y

=3.20 l/(mg s), a

o

=1.2910

ÿ11

mg O

3

/CFU and

b

o

=0.324 mg O

3

/mg C). As depicted in Fig. 4, the

model predicted the data well within the range of

experimental conditions investigated.

The inactivation of E. coli with ozone in the pre-

sence of natural organic matter such as humic acid

can be represented with the integrated form of

equation 1, or

ln



N

N

o



ˆ ÿk

i

…

t

0

c dt

…9†

where c changes with time according to equations

6±8. The results from inactivation experiments per-

formed with solutions containing humic acid at

concentrations TOC

o

=0.1±0.5 mg/l are shown in

Fig. 5 against the corresponding integrated ozone

exposure estimated from predicted curves similar to

those shown in Fig. 4. Also shown in Fig. 5 is the

line corresponding to an inactivation rate constant

of k

i

=138 l/(mg s) in the absence of humic acid esti-

Fig. 4. Apparent overall decomposition of ozone in the

simultaneous presence of initially viable E. coli and humic

acids for various initial ozone and total organic carbon

concentrations, and micro-organism densities (see set III

in Table 1 for experimental conditions).

Fig. 3. Apparent overall decomposition of ozone in the

presence of humic acids for various initial ozone and total

organic carbon concentrations (see set II in Table 1 for ex-

perimental conditions).

Inactivation of E. coli with ozone

2637

mated with equation 2 for the experimental tem-

perature of 208C (Hunt and MarinÄas, 1997). In gen-

eral, good agreement is observed between the

inactivation data in the presence of humic acid and

the line corresponding to the absence of natural or-

ganic matter.

A relationship can now be established between

the inactivation kinetics and the kinetics corre-

sponding to the reaction between molecular ozone

and the fast-reacting portion of E. coli cells:



N

N

o



ˆ

 x

x

o



k

i



k

x

…10†

Equation 10 was obtained by combining equation

9 with an expression of the same form resulting

from integrating equation 3. Values for the corre-

lation between N/N

o

and x/x

o

according to

equation 10 are presented in Table 2 together with

the corresponding predicted ozone exposure.

Approximately 2.8% loss of the fast-reacting con-

stituents of E. coli corresponded to 99% inacti-

vation of this micro-organism and by the time 25%

of the fast-reacting groups had reacted with molecu-

lar ozone, practically all micro-organisms present in

the solutions tested in this study (N

o

110

10

CFU/l)

were inactivated.

Representative micrographs obtained for the

TEM analyses are shown in Fig. 6. Corresponding

ozone initial concentration and exposure, exper-

imental and model predicted E. coli survival and

fast-ozone demand constituent residual fraction are

presented in Table 3. Cells representative of those

present in control samples not exposed to ozone are

shown in micrograph A. The various shapes shown

in the micrograph resulted from random orientation

of the rod-shaped E. coli cells embedded in the

epoxy resin. The lighter part of the cells was the

nuclear region containing some DNA ®brils and

electron-dense ribosomes, the latter appearing as

dark black dots (VanDemark and Batzing, 1987).

Cell capsules and outer membranes were easily dis-

tinguished but no distinct internal membrane was

visible. These observations were consistent with

those reported by Woldringh and Nanninga (1985)

for E. coli also ®xed ®rst with 2.5% glutaraldehyde

and then with 1% osmium tetroxide. These authors

indicated that this ®xation approach resulted in a

nucleoid with preserved disperse appearance with-

out the excessive precipitation of DNA ®brils result-

ing from the use of other ®xation solutions.

Micrograph B presents a sample treated for 30 s

with an initial ozone concentration of 9 mg/l. The

model described previously was applied to predict a

corresponding integrated exposure to ozone of

0.0826 mg s/l, an inactivation of approximately

99.999% and a fast-reacting group conversion of

13.2%. The predicted inactivation matched closely

the observed value. As shown in Fig. 6, the appear-

ance of cells in micrograph B, most of which have

lost viability, was not noticeably di€erent from that

of the control.

Micrograph C in Fig. 6 corresponded to a sample

exposed to 0.175 mg s/l according to model predic-

tions. The model predicted survival ratio presented

in Table 3 should be considered more representative

than that observed which corresponded to approxi-

mately the detection limit of viable E. coli cells and

might have been a€ected by the occurrence of cell

clumps. Essentially all cells exposed to the con-

ditions corresponding to micrograph C were inacti-

vated. Compared to the two previous micrographs,

distinct changes were observed in the bacterial cells.

The nucleoid seemed to have contracted and a

coarse precipitation of DNA was apparent. An

e€ect similar to that observed in micrograph C has

been reported (Woldringh and Nanninga, 1985)

when the secondary structure of the DNA-binding

proteins is destroyed. Apparently, for the level ex-

posure corresponding to micrograph C, ozone was

Fig. 5. Inactivation of E. coli with ozone in the presence

of humic acids for various initial ozone and total organic

carbon concentrations and micro-organism densities (see

set IV in Table 1 for experimental conditions).

Table 2. Model prediction of micro-organism survival, fast ozone-

demand components residual, and corresponding integrated pro-

duct of ozone concentration and contact time

N/N

o

x/x

o

„

t

0

c dt

1

1

0

0.5

0.991

0.00502

10

ÿ1

0.972

0.0167

10

ÿ2

0.944

0.0334

10

ÿ3

0.917

0.0501

10

ÿ4

0.891

0.0667

10

ÿ5

0.866

0.0834

10

ÿ6

0.841

0.100

10

ÿ7

0.818

0.117

10

ÿ8

0.794

0.133

10

ÿ9

0.772

0.150

10

ÿ10

0.750

0.167

Nimrata K. Hunt and Benito J. MarinÄas

2638

Fig. 6. Transmission elctron micrographs of E. coli cells after exposure to ozone in a continuous-¯ow tubular reactor.

Operating conditions: pH 7.2; temperature=208C; N

o

=510

9

CFU/l; t=30 s; c

o

=0 mg/l (A), 9 mg/l (B), 18 mg/l (C) and

196 mg/l (D).

Table 3. Model prediction of micro-organism survival, fast ozone-demand components residual, and corresponding integrated product of

ozone concentration and contact time for the samples investigated by transmission electron microscopy (see Fig. 6)

Micrograph in Fig. 6

c

o

(mg/l)

„

t

0

c dt (mg s/l)$

N/N

o

x/x

o

observed

predicted

A

0

0

1

1

1

B

9

0.0826

110

ÿ5

1.1210

ÿ5

0.868

C

18

0.175

110

ÿ7

3.2510

ÿ11

0.740

D

197

4.21

<10

ÿ7

%

110

ÿ252

0.00071

E}

1000

28.1

<10

ÿ7

%

±%

10

ÿ21

$

Contact time t=30 s for all experiments.

%

No colonies formed.

}

Not shown in Fig. 6. A pellet could not be obtained by centrifugation.

Inactivation of E. coli with ozone

2639

able to penetrate the cells, and react with the pro-

teins, or the numerous enzymes involved in the con-

trol of DNA conformation in the nucleoid,

resulting in the precipitation of DNA. Another

major di€erence between micrograph C and the two

previous pictures was the occurrence of more pro-

nounced convolutions of the cell envelope.

Micrograph D in Fig. 6 illustrates bacterial cells

after a predicted exposure to ozone of 4.21 mg s/l.

The overall density of the nuclear material was low

compared to the previous micrographs and the cell

envelopes appeared severely convoluted and hang-

ing loosely around the cell instead of ®tting snugly.

Some of the bacteria appeared to be disrupted, and

fragments of the lysed cells were observed. The

occurrence of cell lysis was further supported by the

fact that a pellet could not be obtained for a sus-

pension exposed to 28.1 mg s/l (sample E in Table

3).

The rate of E. coli inactivation with ozone is fast

compared to those for other micro-organisms of

public health interest such as protozoan parasites,

e.g. Cryptosporidium parvum. Drinking surface

water systems designed to inactivate C. parvum

oocysts must provide CT values several thousand

times greater than those required to protect against

E. coli (Rennecker et al., 1999). Consequently, the

E. coli inactivation kinetics developed in this study

might not have a direct impact on the design of

drinking water treatment plants. However, this in-

formation is of greater relevance in wastewater

treatment. For example, secondary wastewater e‚u-

ents from the Belmont and Southport Advanced

Wastewater Treatment Plants at Indianapolis, IN

are disinfected with ozone for the primary regulat-

ory compliance goal of reducing the density of

viable fecal coliforms (Blank et al., 1993).

Reductions in fecal coliforms achieved in the ozone

contactors of these plants are reported at 99±

99.99%, or N/N

o

=0.0001±0.01 (Blank et al., 1993).

These inactivation eciencies are low despite these

full-scale disinfection contactors being designed for

operation at applied ozone dosages of 6 mg/l and

t

10

contact time (time for 10% of tracer mass added

as a pulse-input immediately up-stream from the

contactor inlet to reach the e‚uent) of 10 min

(Blank et al., 1993). Assuming that the average

fecal coliform resistance to ozone was the same as

that found in this investigation for E. coli and

neglecting the presence of bacterial clumps then

Fig. 5 can be used to estimate the CT values corre-

sponding to the observed N/N

o

at 0.03±0.07 mg s/l.

The average dissolved ozone concentrations corre-

sponding to these CT values can be estimated at

0.5±110

ÿ4

mg/l assuming that t

10

provided a good

estimate for the e€ective contact time achieved in

the contactors. As illustrated with this example, the

eciency of wastewater e‚uent disinfection with

ozone is a€ected by ozone-demand reactions. If the

apparent decay rate of ozone in a certain waste-

water e‚uent is known then a kinetic model similar

to that developed for this study could be integrated

with reactor mass transfer and hydrodynamic infor-

mation to predict the performance and perhaps

optimize the design of ozone disinfection units.

CONCLUSIONS

The apparent decomposition of dissolved ozone

in the presence of humic acid and E. coli cells was

modeled successfully with mixed second-order rate

expressions within a time scale relevant to E. coli

inactivation with ozone. Kinetic parameters

obtained for the separate reactions between ozone

and humic acid, and ozone and E. coli cells pre-

dicted well the overall decomposition of ozone in

the simultaneous presence of humic acid and E. coli

cells.

The rate for the ozone inactivation of E. coli with

ozone in the presence of humic acid was slower

than that in the absence of natural organic matter.

The di€erence was due to the faster decomposition

of dissolved ozone and, thus, the lower exposure of

E. coli cells to the disinfectant. However, the sec-

ond-order inactivation rate constant was approxi-

mately the same in the presence and absence of

humic acid con®rming that dissolved ozone is the

species generally responsible for inactivation.

The overall chemical reaction between ozone and

organic matter in E. coli cells was slower that the

corresponding inactivation reaction. Only 25% of

the initial relatively fast ozone demand were satis-

®ed by the time that practically all micro-organisms

present in solution were inactivated. TEM micro-

graphs revealed that noticeable changes in the in-

terior of E. coli cells did not take place until most

of the cells present in the sample were non-viable.

Subsequent exposure to ozone resulted in structural

changes, membrane deterioration, and ultimately

lysis of the inactivated cells.

AcknowledgementsÐThe authors would like to thank Dr

Bruce Hunt, Dr Chung-Fan Chiou, Manalee Johnson and

Tom Cooper, School of Civil Engineering, Purdue

University, West Lafayette, IN for their technical assist-

ance and support at various stages of this study. TEM

analyses were performed by Dr John Turek and the lab-

oratory technicians at the School of Veterinary Medicine,

Purdue University, West Lafayette, IN. Financial support

by the City of Indianapolis, Department of Public Works,

Indianapolis, IN and the Schools of Engineering, Purdue

University, West Lafayette, IN are gratefully acknowl-

edged.

REFERENCES

Blank B. D., MarinÄas B. J., Corsaro K. and Rakness K.

L. (1993) Enhancement of wastewater disinfection e-

ciency in full-scale ozone bubble-di€user contactors.

Ozone Science and Engineering 15, 295±310.

Chick H. (1908) An investigation of the laws of disinfec-

tion. Journal of Hygiene 8, 92±158.

Nimrata K. Hunt and Benito J. MarinÄas

2640

Christensen E. and Giese A. C. (1954) Changes in absorp-

tion spectra of nucleic acids and their derivatives follow-

ing exposure to ozone and ultraviolet radiations.

Archives of Biochemistry and Biophysics 50, 206±209.

Hamelin C. and Chung Y. S. (1974) Optimal conditions

for mutagenesis by ozone in Escherichia coli K12.

Mutation Research 24, 271±279.

Hamelin C., Sarhan F. and Chung Y. S. (1977) Ozone

induced DNA degradation in di€erent DNA polymerase

I mutants of Escherichia coli K12. Biochemical and

Biophysical Research Communication 77, 220±224.

Hamelin C., Sarhan F. and Chung Y. S. (1978) Induction

of deoxyribonucleic acid degradation in Escherichia coli

by ozone. Experentia 34, 1578±1579.

Hunt N. K. and MarinÄas B. J. (1997) Escherichia coli inac-

tivation with ozone. Water Research 31, 1355±1362.

Ishizaki K., Miura K. and Shinriki N. (1987) E€ect of

ozone on plasmid DNA of Escherichia coli in situ.

Water Research 21, 823±827.

Perich J. R., McCammon L. R., Cronholm L. S.,

Fleischman M., Pavoni J. L. and Riesser V. (1975)

Inactivation kinetics of virus and bacteria in a model

ozone contacting reactor system. In Proceedings of the

Second International Symposium on Ozone Technology,

Montreal Canada, eds. R. G. Rice, P. Pichet and M.

Vincent, p. 486. International Ozone Institute, New

York, NY.

Pryor W. A., Dooley M. M. and Church D. F. (1983)

Mechanisms of the reactions of ozone with biological

molecules: the source of toxic e€ects of ozone. In

Advances in Modern Environmental Toxicology, eds. M.

G. Mustafa and M. A. Mehlman. Ann Arbor Science

Publishers, Ann Arbor, MI.

Rennecker J. L., MarinÄas B. J., Owens J. H. and Rice E.

W. (1999) Inactivation of Cryptosporidium parvum

oocysts with ozone. Water Research 33(11), 2481±2488.

Scott D. B. M. (1975) The e€ect of ozone on nucleic acids

and their derivatives. In Aquatic Application of Ozone,

eds. W. J. Blogoslawski and R. G. Rice, p. 1.

International Ozone Institute, New York, NY.

Scott D. B. M. and Lesher E. C. (1963) E€ect of ozone on

survival and permeability of Escherichia coli. Journal of

Bacteriology 85, 567±576.

VanDemark P. J. and Batzing B. L. (1987) The Microbes:

an Introduction to their Nature and Importance.

Benjamin/Cummings Publishing Company, Menlo Park,

CA.

Watson H. E. (1908) A note on the variation of rate of

disinfection with the change in the concentration of dis-

infectant. Journal of Hygiene 8, 536±542.

Woldringh C. L. and Nanninga N. (1985) Nucleoid and

cytoplasm in the intact cell. In Molecular Cytology of

Escherichia coli, ed. N. Nanninga. Academic Press,

London, UK.

Zhou H. and Smith D. W. (1994) Kinetics of ozone disin-

fection in completely mixed system. J. Environ. Eng.

Div. ASCE 120, 841±858.

Inactivation of E. coli with ozone

2641