ARTICLE IN PRESS

Water Research 39 (2005) 3098–3108

www.elsevier.com/locate/watres

Comparison of electrocoagulation and chemical

coagulation pretreatment for enhanced virus removal using
microfiltration membranes
Bintuan Zhua, Dennis A. Clifforda,, Shankararaman Chellama,b
a
Department of Civil and Environmental Engineering, University of Houston, 4800 Calhoun Road, Houston, TX 77204-4003, USA
b
Department of Chemical Engineering, University of Houston, Houston, TX 77204-4004, USA
Received 20 December 2004; received in revised form 17 May 2005; accepted 18 May 2005

Abstract

This research studied virus removal by iron electrocoagulation (EC) followed by microfiltration (MF) in water
treatment using the MS2 bacteriophage as a tracer virus. In the absence of EC, MF alone achieved less than a 0.5-log
removal of MS2 virus, but, as the iron-coagulant dosage increased, the log virus removal increased dramatically. More
than 4-log virus removal, as required by the Surface Water Treatment Rule, was achieved with 6–9 mg/L Fe3+. The
experimental data indicated that at lower iron dosages and pH (o
8 mg Fe/L and pH 6.3 and 7.3) negatively charged
MS2 viruses first adsorbed onto the positively charged iron hydroxide floc particles before being removed by MF. At
higher iron dosages and pH (4
9 mg Fe/L and pH 8.3), virus removal was attributed predominantly to enmeshment
and subsequent removal by MF. Additionally, the experimental data showed no obvious influence of ionic strength in
the natural water range of 107–102 M on MS2 virus removal by EC-MF. Finally, EC pretreatment significantly
outperformed chemical coagulation pretreatment for virus removal. The proposed mechanism for this improved
performance by EC is that locally higher iron and virus concentrations and locally lower pH near the anode improved
MS2 enmeshment by iron flocs as well as adsorption of MS2 viruses onto the iron floc particles.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Microfiltration; Coagulation; Electrocoagulation; Virus; Bacteriophage; Water treatment; Membrane filtration

1. Introduction Several published studies have reported partial (0.2- to

Microfiltration (MF) has been used to completely or
significantly remove turbidity, bacteria, and protozoa
from water and wastewater (Jacangelo et al., 1991;
Mallevialle et al., 1996; Madaeni, 1999). However, MF
alone is not an efficient barrier for virus removal,
because viruses are typically smaller than its pores.
Corresponding author. Tel.: +713 743 4266;
fax: +713 743 4260.
E-mail address: daclifford@uh.edu (D.A. Clifford).
3-log) virus removal by MF (Coffey et al., 1993;
Madaeni et al., 1995; Urase et al., 1996; Roberts,
1997), which is significantly less than the 4-log virus
removal mandated by the Surface Water Treatment
Rule (SWTR). Therefore, MF by itself cannot meet the
SWTR virus removal requirement even though the
turbidity and protozoa regulations can be met easily.
The addition of a coagulant prior to membrane
filtration has been suggested to generally improve
product water quality and reduce membrane fouling
by coagulating the dissolved organic matter ahead of the

0043-1354/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2005.05.020
 ARTICLE IN PRESS
B. Zhu et al. / Water Research 39 (2005) 3098–3108
filter (Lahoussine-Turcaud et al., 1990; Wiesner et al.,
1992). Usually, chemical coagulation (CC) with iron or
aluminum salt is used, however electrocoagulation (EC)
is another possibility. To our knowledge, this is the first
report of EC pretreatment for MF.
EC has been widely studied in water and wastewater
treatment to remove heavy metals, organics, bacteria,
hardness, turbidity, and other contaminants (Horner
and Duffey, 1983; Tsouris et al., 2001; Can et al., 2003;
Mills, 2000). In the EC process, the electrodes are
consumed as the coagulant is generated and precipi-
tated; no liquid chemical is added; alkalinity is not
consumed; and pH adjustment is not needed. Addition-
ally, compared with CC, the EC process reportedly
requires less coagulant and produces less sludge (Horner
and Duffey, 1983; Mills, 2000). According to one
estimate, the space required for EC is less than CC
because EC does not require chemical storage, dilution,
and rapid mixing (Mills, 2000). Because EC systems
typically use solid iron or aluminum anodes rather than
corrosive iron or aluminum salt solutions, EC units can
be more easily incorporated into ‘‘packaged’’ plants and
transportable water treatment plants for use in remote
areas or in emergency water supply treatment, which
was one of the driving forces to undertake this research.
The objectives of this study were to (a) evaluate the
efficacy of iron EC-MF for virus removal, (b) investigate
the effect of EC-MF operating parameters on virus
removal, (c) compare virus removal by EC-MF and CC-
MF, and (d) suggest possible mechanisms for virus
removal by EC-MF.
2. Materials and methods
2.1. Virus
The indicator-organism virus used throughout the
study was the bacterial virus, MS2. It is 0.025 mm in size,
icosahedral in shape, and contains a single strand of
ribonucleic acid with 3569 nucleotides (Valegård et al.,
1990). The virus stock was bought from American Type
Culture Collection (ATCC # 15597-B1). Because bac-
terial viruses are also called bacteriophages (Brock and
Madigan, 1991), we use the terms ‘‘virus’’ and ‘‘bacter-
iophage’’ interchangeably to refer to MS2.
2.2. MS2 virus propagation and MS2 virus assay
The procedure of MS2 virus propagation and MS2
virus assay was described in our previous paper (Zhu et
al., 2004). MS2 samples were assayed using a modified
agar-overlay technique, which counts viable viruses
using Escherichia coli as the host bacterium.
3099
2.3. Test water
Three synthetic fresh waters, roughly characterized as
low (I ¼ 107 M), medium (I ¼ 6  103 M), and high
ionic strength (I ¼ 1:8  102 M), were used in this
research. These were prepared from deionized (DI)
water and reagent-grade chemicals to simulate fresh
water containing monovalent and divalent cations, and
alkalinity. The low-ionic-strength water was DI water,
and the high-ionic-strength water was DI water spiked
with 3.0 mM NaHCO3, and 10 mM CaCl2. The medium-
ionic-strength water used in most of the experiments
consisted of DI water spiked with 3.0 mM NaHCO3, and
1.0 mM CaCl2. Its pH was 8.3 without any adjustments.
2.4. Membrane microfilter
All filtration experiments were carried out using a
hydrophilic, 0.22-mm pore size, modified polyvinylidene
fluoride (PVDF) membrane filter (Durapore, Millipore
Corp. Bedford, MA). A fresh membrane with an
effective filtration area of 4.1 cm2 was used for each
experiment. The manufacturer confirmed that its surface
was negatively charged in the pH range of our
experiments, 6.3–8.3.
2.5. EC unit
A bench-scale EC unit with three anode–cathode pairs
was designed and built for this research (Fig. 1). It
consisted of a 200-mL active-volume, flow-through
electrode chamber with 22-cm long rod-shaped iron
anodes and porous cylindrical stainless steel cathodes.
The total anode surface area was 100 cm2 and the
current density was typically 0.25 mA/cm2. By adjusting
the operating current and flow rate of source water, the
desired iron concentration was obtained.
As can be seen in Fig. 1, the recycle pump discharged
into the annular area between the cathode and anode of
each of the three cathode-anode pairs in order to flush
the anode where iron coagulant was continuously
Source Water
PlanView of Anode and Cathode
Internal Diameter: 0.85 cm
EC Unit
Stainless Steel Cathode
Iron Anode Diameter: 0.48 cm
Recycle Pump
Fig. 1. Diagram of batch bench-scale EC system.
 ARTICLE IN PRESS
3100 B. Zhu et al. / Water Research 39 (2005) 3098–3108
generated. During the continuous flow experiments (see
Section 2.8), raw water entered the unit through an inlet
tube that terminated near the center of the unit at about
one half the water depth.
2.6. Calculation of virus removal
As is commonly practiced, virus removal in these
experiments is expressed as a log reduction value (LRV)
calculated using the permeate (CP) and feed concentra-
tions (CF) of the MS2 virus (EPA, 2001):
LRV ¼ log removal value ¼ logðC F =C P Þ. (1)
Because the target virus concentration in the feed
water was 106 pfu/mL, the maximum achievable LRV in
our experiments was 6.0. (Experimental LRVs were
lower than 6.0 in all cases.)
2.7. Fe (III) measurement
The concentrations of Fe3+ generated by the EC unit
were analyzed in triplicate using flame atomic absorp-
tion spectroscopy (Flame AA-AAnalyst 300, Perkin-
Elmer Corporation, CT, USA) according to Method
3111 in Standard Methods (Clesceri et al., 1998).
2.8. Bench-scale EC-MF system
The EC unit treated source water containing viruses
before it was microfiltered. Source water entered from
the top of the EC unit. Once inside the unit, it was
recycled at about 500 mL/min to ensure that the
electrodes remained free of surface deposits and that
the Fe3+ generated was quickly released from the
surface and hydrolyzed. Following the practice of other
EC researchers (Tsouris et al., 2001; Han et al., 2002;
Holt et al., 2002), all virus-removal data from the EC
unit was collected in a batch mode, i.e., the EC unit was
filled with virus-spiked water, and turned on for 40 s to
generate the desired target amount of iron. After EC, the
solution was directly filtered by transferring it to the feed
bottle where it was gently and continuously mixed
during MF. The bench-scale membrane system was
operated in the dead-end filtration mode at a constant
transmembrane pressure of 69 kPa (10 psig) maintained
by compressed air. Finally, samples for analysis of
viruses were taken before the EC unit, after the EC unit,
and after the membrane filter.
Although batch EC experiments were carried out
under non-steady-state conditions as was the practice of
other researchers, a limited number of continuous flow,
steady-state EC-MF experiments for virus removal were
also carried out at pH 7.3 with varying iron dosages
using the same source water as used in the batch
experiments. The procedure used for the continuous
experiments was as follows: 300 mL/min of source water
spiked with MS2 virus was fed into the top of empty EC
unit and when the level reached the overflow port, the
water was recycled at about 500 mL/min. Prior to
reaching steady state, about 2 L of effluent water was
wasted. After reaching steady state as predetermined by
the volume required reach a constant iron concentration
in the effluent, the effluent water was directed to the feed
bottle of the MF system. Viruses in the EC unit feed
tank and in the MF effluent were counted for the
determination of log removal using Eq. (1).
2.9. Bench-scale CC-MF system
In order to compare EC-MF with chemical coagula-
tion (CC)-MF, CC-MF experiments were carried out
using the procedure described in a previous paper (Zhu
et al., 2004). Source water spiked with MS2 virus was
coagulated using FeCl3 (1 min rapid mixing at 100 rpm),
flocculated (30-min slow mixing at 30 rpm), and then
filtered in the dead-end mode using the PVDF flat-sheet
membrane.
2.10. MS2 viability during the course of the experiments
If virus inactivation occurred during EC pretreatment,
it would result in an overestimation of intrinsic log virus
removal by combined EC and membrane filtration. To
determine if virus inactivation occurred as a result of
EC, pressure change, or mixing during MF, two
separate control experiments were conducted. The first
involved EC only and the second involved pressure and
mixing during filtration. In the first experiment, viruses
were electrocoagulated, and then samples were taken
from the EC unit for viral counting without performing
any filtration.
In the second experiment, which determined the effect
of pressure and mixing on virus viability, the initial
sample (Sample 1) for virus enumeration was taken from
the source water before EC; the final sample (Sample 2)
was taken from the last 20 mL of coagulated feed water
present in the pressurized (69 kPa, 10 p) feed bottle. This
sample (Sample 2) had been electrocoagulated, pressur-
ized and mixed for about 1 h (no filtration was
performed). The difference in viral counts between
Samples 1 and 2 was considered to be a measure of
the effects of pressure and mixing on viability.
2.11. Bench-scale CC-MF system using the EC unit as a
hydraulic mixer only
During the EC experiments, water in the EC unit was
re-circulated at a high rate (500 mL/min) to keep it well-
mixed. Such hydraulic mixing differed from the mechan-
ical mixing provided by a rectangular paddle mixer
during chemical coagulation jar tests, and could possibly
have changed virus removal by MF. Therefore, to insure
 ARTICLE IN PRESS
B. Zhu et al. / Water Research 39 (2005) 3098–3108
that any differences between EC and CC could be
attributed solely to the electrochemical generation of
Fe3+, chemical coagulation was conducted in the EC
unit with the power to the electrodes turned off and by
using recirculation for 40s rapid mixing during chemical
coagulation.
The source water in these CC control experiments
using the EC unit for mixing had the same composition
as that used in the EC-MF experiments. Four coagulant
dosages—0, 2, 5, and 10 mg Fe/L—were targeted. The
procedure for chemical coagulation using the EC unit
solely to provide hydraulic mixing was as follows: (a)
400 mL of source water containing MS2 viruses was
added from the top of the EC unit using a glass beaker;
(b) simultaneously, a predetermined dose of FeCl3 stock
solution (2000 mg-Fe/L) was quickly added to the EC
unit from the top using a syringe; and (c) the recycle
pump (500 mL/min) was turned on while the EC power
supply remained off. The detention time of the source
recirculating water in the un-powered EC unit was
maintained at 40 s, which was the same as all other EC-
MF experiments. After 40 s of recycle mixing in the un-
powered EC unit, the chemically coagulated solution
was immediately transferred to the feed bottle where it
was gently and continuously mixed during filtration.
The operation of the bench-scale membrane system
following chemical coagulation was the same as that
described in the bench-scale EC-MF system. Finally,
samples for virus analysis were taken before coagulation
and after MF.
3. Results and discussion
3.1. Operation of EC unit
After designing and fabricating the EC unit, it was
tested for its ability to generate iron in proportion to the
current by operating it continuously at different current
values, while source water was pumped into the top at a
constant flow rate of 300 mL/min. Water was recycled in
the unit at 500 mL/min and exited through the overflow
port at 300 mL/min. Samples for analysis of Fe(OH)3(s)
measured as Fe, generated by EC unit, were taken from
the effluent of the EC unit at each different operating-
current condition.
To determine the current efficiency, the amount of
iron generated was calculated using Faraday’s Law (Eq.
(2)):
I  t  MW
m¼ , (2)
ZF
where m is the mass in grams of Fe generated at a
specific current (I, amperes) over a time interval
(t, seconds), Z is the number of electrons transferred
per Fe atom, MW is the molecular weight
3101
(55.85 g mol1), and F is Faraday’s constant (96,486 C
eq1).
When operated at steady state, which was reached
after approximately 8 EC unit volumes (1600 mL) of
throughput, the Fe3+ concentration (mg/L) in the EC
effluent water was a function of the current and the
detention time in the EC unit. By adjusting the operating
current and flow rate of source water to be treated, a
desired iron concentration was obtained. For example,
when the feed flow rate was 300 mL/min and the
operating current was 0.25 A, the steady-state iron
concentration was 9.6 mg/L. Based on the observed
steady state Fe(III) concentration and operating current,
the EC unit operating curve shown in Fig. 2 was
obtained. The calculated current efficiency was 93%
based on the slope of the linear best-fit curve with zero
intercept.
In all cases, no iron was detected (method detection
limit 0.03 mg Fe/L) in the microfiltered water indicating
that the EC unit predominantly generated particulate
iron.
3.2. Possibility of MS2 virus removal or inactivation
within the EC unit
Our previous research demonstrated that chemical
coagulation with ferric chloride did not inactivate MS2
viruses (Zhu et al., 2004). However, because EC uses
voltage and current to generate iron, chlorine produc-
tion is also possible, which would inactivate viruses (Liu
et al., 1997; Drees et al., 2003).
Experiments were carried out to investigate the
possibility of inadvertent virus inactivation by chemical
disinfection or virus removal by accumulation on the
anode. Virus samples were taken from the spiked raw
water (Sample 1) and the electrocoagulated water
immediately after operating the EC unit (Sample 2). If
12
Fe(III) Concentration(mg/L)
10
8
6
4
Experimental Data
2 Theoretical Value
0
0 0.05 0.1 0.15 0.2 0.25 0.3
EC Operating Current (A)
Fig. 2. Fe(III) generated at 300 mL/min flow rate as a function
of EC operating current.
 ARTICLE IN PRESS
3102 B. Zhu et al. / Water Research 39 (2005) 3098–3108
chlorine was generated or viruses were accumulated on
the EC anode, a reduction in the number of viable
viruses would be observed. The results of triplicate
experiments conducted at different iron doses and pH
values are shown in Fig. 3. The virus concentrations of
Samples 1 and 2 were compared using the t-test based on
n2 ¼ 4 degrees of freedom and a 0.05 level of
significance. All test statistics were less than the critical
value (t0.05(4) ¼ 2.776). Hence, the virus concentrations
before and after EC treatment were essentially the same
demonstrating that viruses did not accumulate on the
anode nor were viruses rendered less viable by EC.
3.3. Effect of mixing- and pressurization on MS2 virus
viability during filtration
The results of additional control experiments con-
ducted to establish that virus concentrations in the
electrocoagulated mixture in the pressurized feed bottle
did not change significantly during MF are shown in
Fig. 4. Sample 1 (N 0 ) was taken before the EC unit, and
Sample 2 (N) was taken from the feed bottle near the
end of filtration. The virus concentrations of Sample 1
(average of 3 replicates) and Sample 2 (average of 3
replicates) were also compared using the t-test.
Samples 1 and 2 were not significantly different based
on the t-test statistics, all of which were less than critical
value of t0.05(4) ¼ 2.776. Therefore, operating pressure
(69 kPa, 10 psig) or stirring time (
1 h) had no effect on
virus viability demonstrating that any virus removal
observed during EC-MF experiments was due to the
combination of EC and filtration and not inactivation
during extended mixing at 69 kPa (10 psig). These results
are consistent with those in our previous CC-MF study
(Zhu et al., 2004).
2.0
Normalized MS2 Virus
Concentration (N/N0)
1.5
1.0
0.5
pH =6.3
pH =7.3
pH =8.3
0.0
0 3 6 9 12
Fe(III) Concentration (mg/L)
Fig. 3. Determination of virus inactivation by electrocoagula-
tion. (N is MS2 concentration after EC treatment and N 0 is
MS2 concentration before EC treatment.) Error bars represent
one standard deviation.
2.0
Normalized MS2 Virus
Concentration (N/N0)
1.5
1.0
0.5
pH = 6.3
pH = 7.3
pH = 8.3
0.0
0 2 4 6 8 10 12 14
Fe(III) Concentration (mg/L)
Fig. 4. Determination of mixing and pressurization on MS2
viability during filtration. Error bars represent one standard
deviation.
3.4. Effect of Fe (III) concentration and cake thickness
on MS2 virus removal by EC-MF
The results of EC-MF experiments conducted to
determine the effects of iron dosage and pH on virus
removal in medium ionic-strength background water
with 106 pfu/mL virus concentration are shown Fig. 5.
In the absence of EC, MF alone achieved less than 0.5-
log virus removal. With increasing iron dosage during
EC pretreatment, virus removal was dramatically
improved. The 4-log virus removal required by the
SWTR (EPA, 2001) was achieved at Fe(III) concentra-
tions in the range of 6–9 mg/L as pH increased from 6.3
to 8.3.
The membrane alone (220 nm nominal pore size) was
not expected to substantially remove the 25 nm MS2
bacteriophage directly by sieve action. Moreover,
because the MS2 bacteriophage (isoelectric pH, 3.9)
and the surface of the membrane are both negatively
charged in the studied pH range (6.3–8.3), substantial
virus adsorption on the membrane was not expected.
The significant virus reductions shown in Fig. 5 suggests
two mechanistic possibilities: (1) enmeshment of viruses
by the Fe(OH)3 precipitates, and/or (2) a two-step
process where initially Fe(OH)3 flocs are formed, which
are positively charged in the pH range 6.3 and 7.3 and
near neutral at pH 8.3 (O’Melia, 1972). In the second
step, the negatively charged viruses adsorb onto the
positively charged Fe(OH)3(s) flocs that were subse-
15 quently well removed by MF. A similar mechanism
applies for adsorption of soluble anions onto hydrous
metal oxides such as goethite (FeOOH) and gibbsite
(AlOOH) (Hingston, 1981; Schindler, 1981). This is also
analogous to the viruses concentration procedure
recommended in ‘‘Standard Method 9510D’’ (Clesceri
 ARTICLE IN PRESS
B. Zhu et al. / Water Research 39 (2005) 3098–3108
6 6
5 dominant dominant
4
pH = 8.3
3 4
Fe(III) = 12.6 mg/L
Fe(III) = 9.8 mg/L
2 Fe(III) = 7.4 mg/L
Fe(III) = 4.5 mg/L
1 Fe(III) = 2.3 mg/L
Log Removal of MS2 Bacteriophage
Fe(III) = 0.0 mg/L
0 pH=6.3
5 pH=8.3
4
pH = 7.3
3 Fe(III) Concentration (mg/L)
Fe(III) = 13.1mg/L
Fe(III) = 10.1mg/L
2 Fe(III) = 7.3 mg/L
Fe(III) = 6.4 mg/L
1 Fe(III) = 2.8 mg/L
Fe(III) = 0.0 mg/L
0
5
4 pH = 6.3
3 Fe(III) = 10.0 mg/L
Fe(III) = 7.6 mg/L
Fe(III) = 6.0 mg/L
2 Fe(III) = 4.9 mg/L
Fe(III) = 3.0 mg/L
1 Fe(III) = 1.7 mg/L
Fe(III) = 0.0 mg/L
0 Fe(III) dosages. At pH 6.3, Fe(OH)+ 2 and Fe(OH)
0 20 40 60 80 100 120
Volume Filtered/Membrane Area (mL/cm2)
Fig. 5. The effect of coagulant dose and volume filtered (cake
thickness) on MS2 removal by EC-MF.
et al., 1998) wherein they are first adsorbed onto
positively charged flocs and later eluted in alkaline
solution where flocs become negatively charged and/or
more soluble.
Log virus removal increased with iron-coagulant dose
(Fig. 5) because as dose increased, (1) the coagulant
surface area and number of adsorptions sites available
increased and (2) more Fe(OH)3 precipitated that
encapsulated additional viruses. The above mechanisms
of virus removal are consistent with the negligible
impact of membrane cake layer thickness on virus
removal, which can be deduced from Fig. 5, where cake
layer thickness increased with increasing volume filtered.
As observed, virus removal remained approximately
constant at any fixed iron dose even as thicker cake
layers were deposited as filtration progressed. These
results are similar to those observed in our previous iron
coagulation-microfiltration research (Zhu et al., 2004).
To further investigate virus removal mechanisms,
experimental data from different pH conditions were
examined.
3103
Log Removal of MS2 Bacteriophage
Adsorption Enmeshment
5
SWTR virus requirement
3
2
1 pH=7.3
0
0 2 4 6 8 10 12 14
Fig. 6. The effect of iron dose and pH on MS2 log removal by
EC-MF.
3.5. Effect of pH on MS2 virus removal by EC-MF
As shown in Fig. 6, a bulk solution pH of 6.3 and
lower Fe(III) dosages (o
9 mg/L) generally achieved
slightly higher MS2 removal compared with 7.3 and
even 8.3, which is approximately the pH of least
solubility for iron (O’Melia, 1972). However, pH did
not have a substantial effect on virus removal at larger
2+
are
the dominant dissolved iron species and are thought to
be the species present on the surfaces of the positively
charged Fe(III) flocs (O’Melia, 1972). At pH 7.3, Fe(III)
flocs are less positively charged compared with pH 6.3
because Fe(OH)+ 2 and Fe(OH)03 are the dominant
dissolved and surface bound iron species. Fe(III) flocs
are near neutral at pH 8.3 because Fe(OH)03 is the
dominant dissolved iron species. Since pKa of the MS2
bacteriophage is 3.9, it is substantially negatively
charged at all pH conditions investigated in this study.
Hence, greater MS2 removal at lower Fe(III) dosages
and at pH 6.3 by EC-MF is consistent with MS2
adsorption onto flocs and subsequent removal by the
membrane. Similar results were obtained in CC-MF
experiments conducted by us (Zhu et al., 2004).
During iron EC, the following electrochemical reac-
tions take place:
0 3+ 
Anode: Fe (s) ¼ Fe +3e and
3+ +
Fe +3H2O ¼ Fe(OH)3(s)+3H
 
Cathode: 3H2O+3e ¼ 3OH +32H2
0
Overall: Fe +3H2O ¼ Fe(OH)3(s)+32H2
Thus, ferric ions produced at the anode hydrolyze to
produce protons, thereby reducing the pH in the area
near the anode. The significant advantage of EC over
CC is that the hydroxide ions generated at the cathode
neutralize these protons in the bulk solution eliminating
 ARTICLE IN PRESS
3104 B. Zhu et al. / Water Research 39 (2005) 3098–3108

the consumption of alkalinity during metal-ion hydro-
lysis, which occurs during CC when ferric or aluminum
salts are used. The bulk-solution-pH effect on virus
removal (Fig. 6) would have been decreased by the fact
that virus adsorption onto the coagulant primarily
occurred in the low-pH region near the anode. Here,
the iron (III) hydroxide complexes would be even more
positively charged at all studied bulk solution pH
conditions (6.3–8.3). Thus, the pH near the anode had
a major effect on MS2 virus adsorption onto iron floc,
whereas, the bulk solution pH had a less noticeable
effect.
At high iron dosages (4
9 mg/L in our system),
Fe(OH)3 precipitates encapsulate MS2 viruses regardless
of pH and remove them by enmeshment resulting in
similar removals at pH 6.3, 7.3 and 8.3 (Fig. 6).
3.6. Effect of ionic strength on MS2 removal by EC-MF
of other researchers (Tsouris et al., 2001; Han et al.,
2002; Holt et al., 2002) have been obtained by unsteady-
state operation of the EC unit in batch mode. In order to
establish the equivalence of batch unsteady-state experi-
mental results described thus far and continuous-flow
EC pretreatment, limited steady-state experiments were
also conducted at pH 7.3 corresponding to water
treatment EC practice. Negligible differences between
the two operating modes at pH 7.3 can be deduced from
Fig. 8. Based on these tests, which were run in triplicate
at five different iron dosages, it can be inferred that
operating an EC unit in batch mode, as is typically
employed in lab-scale studies, provides meaningful
results that are also directly applicable to continuous
operation. (It should be noted that this work did not
explicitly consider scaling-up the microfiltration pro-
cess.)

MS2 removal by EC-MF was investigated in the range
of ionic strength expected in fresh waters: low
(I ¼ 107 M), medium (I ¼ 6  103 M), and high
(I ¼ 1:8  102 M). Results of these experiments, which
were carried out at pH 6.3 are depicted in Fig. 7. As
observed, ionic strength in the 104 to 18 mM range did
not significantly influence virus removal. Regardless of
ionic strength, iron generated by EC dramatically
improved the MS2 virus removal from o0.5 log in the
absence of pretreatment to 44 logs at
7 mg Fe/L.
3.7. Comparison of virus removal by EC-MF between
batch unsteady-state and continuous flow steady-state
operation conditions
In actual practice, EC systems are operated continu-
ously ultimately reaching steady-state. However all
results presented thus far in this paper as well as those
Log Removal of MS2 Bacteriophage
3.8. Comparison of chemical- and electrocoagulation
pretreatment
A comparison of the chemical coagulation and EC
pretreatment methodologies for MF at different pHs
and iron dosages is presented in Fig. 9. Fe(III)
coagulation either by EC or CC pretreatment dramati-
cally improved virus removal. Furthermore, regardless
of pH, the Fe(III) dosage required to achieve a given
MS2 virus LRV was always significantly less for EC
compared with CC pretreatment. For example, 4-log
(99.99%) reduction of viruses was readily achieved by
EC with 6–9 mg/L Fe(III) dosage, whereas only 2-log
virus reduction was achieved with CC pretreatment in
this same Fe(III) dosage range. The fact that EC
significantly outperformed chemical coagulation de-
serves explanation.

Log Removal of MS2 Bacteriophage

6 6

5 5

4 4

3 3

2 2
pH =6.3

1 DI Water + 3 mM NaHCO3 + 10 mM CaCl2

1 Batch mode
DI Water + 3 mM NaHCO3 + 1 mMCaCl2
DI Water Continuous flow mode
0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 14
Fe(III) Concentration (mg/L) Fe(III) Concentration (mg/L)

Fig. 7. The effect of Fe(III) dosage and ionic strength on MS2 Fig. 8. MS2 virus removal by EC-MF at batch non steady-state
virus removal by EC-MF. Feed MS2 virus concentra- and continuous steady-state operation conditions at pH ¼ 7.3,
tion ¼ 106 pfu/mL. feed MS2 virus concentration ¼ 106 pfu/mL.
 ARTICLE IN PRESS
B. Zhu et al. / Water Research 39 (2005) 3098–3108 3105

6
Log Removal of MS2 Bacteriophage

pH =8.3 pH =7.3 pH = 6.3

5

4
EC
EC EC
3

2 CC

CC CC
1

0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Fe(III) Concentration (mg/L)

Fig. 9. Comparison of virus removal by EC-MF and CC-MF as a function of iron dose.

Log removal of MS2 bacteriophage

Fig. 9 demonstrates that iron was used more
effectively during EC compared with CC as performed 3 CC-MF using EC as a mixer
CC-MF using a square-jar mixer
in this study. Iron is generated continuously over the
extended area of the anode in EC compared with the
point addition of iron in CC. Hence, freshly precipitated 2
flocs are generated and more effectively dispersed in EC,
which will increase adsorptive removal of viruses as
reported in ‘‘Standard Method 9510D’’ (Clesceri et al.,
1998). As demonstrated in Section 3.2, there was no 1
virus inactivation or accumulation within the EC unit.
However, differences in mixing or materials of construc-
tion between EC-MF and CC-MF processes could have 0
caused EC to perform better than CC. Results from 0 2 4 6 8 10 12
control experiments designed to evaluate these experi- Fe(III) Concentration (mg/L)
mental artifacts are presented next.
Fig. 10. The effect of possible difference in hydraulic mixing on
virus removal by CC-MF at pH ¼ 6.3, and MS2 feed
3.9. CC-MF for MS2 virus removal using the EC Unit as concentration ¼ 106 pfu/mL.
a mixer

The effect of hydraulic mixing in the EC unit on virus

removal was investigated using the EC unit solely as a
hydraulic mixer for CC with the power turned off (no
electrochemical iron generation). Batch CC-MF tests
were conducted with 400 mL of MS2 virus-spiked test
water. This test water and FeCl3 (at doses of 0, 2, 5, and
10 mg Fe/L) were simultaneously input to the EC unit
Fig. 10 depicts very similar trends for virus removal
by CC-MF using the standard square ‘‘gator’’ jar for
mixing and CC-MF using the EC unit as a mixer with
the recirculation pump on and the EC power off. Log
removal of MS2 virus was essentially the same at a fixed
iron dose under both mixing conditions, demonstrating
that hydraulic mixing in the unit did not enhance virus
removal by EC pretreatment compared with CC. Hence,
virus adsorption onto iron flocs or enmeshment by iron
flocs was significantly better during EC compared with
CC resulting in EC outperforming CC as MF pretreat-
ment for virus removal.
3.10. Proposed explanation for why EC-MF
outperformed CC-MF
During chemical coagulation, the pH, viruses, and
iron hydroxides were uniformly dispersed in the jar.
However, during EC, iron coagulant was continuously
generated and released from the anode surface to water.
Hence, the iron concentration decreased as distance
from the anode increased. Also, the pH in the area near
the anode decreased due to the local hydrolysis of Fe3+
ions to produce protons. This has been verified by Chen
et al. (2002) in an EC system that used an aluminum
 ARTICLE IN PRESS
3106 B. Zhu et al. / Water Research 39 (2005) 3098–3108
anode. Additionally, since the MS2 virus was negatively
charged under our experimental conditions (pH 6.3, 7.3,
and 8.3), and the anode was positively charged after
turning on the EC power, negatively charged MS2 virus
could electrophoretically move resulting in a higher
virus concentration near the anode. Further, in the
immediate region around the anode, there was a locally
higher concentration of Fe(OH)3(s) flocs compared with
the average concentration of Fe(OH)3(s) flocs during
chemical coagulation in the jar mixer. Thus, locally
higher coagulant and virus concentrations might have
been the reasons for better performance of EC
compared with CC. These possibilities are examined in
the following two sections by using a simple model of
virus transport and by performing experiments to
determine the effects of virus and iron coagulant
concentrations on virus removal.
3.10.1. Simple model of virus movement in the EC unit
Two-dimensional virus transport in the EC unit; from
the inlet to the outlet (convection dominated) and from
the cathode to the anode (electrophoresis) was consid-
ered. A simple expression for the virus electrophoretic
velocity (v) was developed by equating the Stokes drag
force and the force induced by the imposed electric field
(Zhu, 2004):
QE   100 mL source water without MS2 viruses present.
v¼ 1  eð6pZrt=mÞ , (3)
6pZr
where Q (Coulombs) is the net surface charge of the
MS2 virus (647, 680, and 685 e at pH 6.3, 7.3, and 8.3,
respectively (Penrod et al., 1996)); E is the electrical field
strength (0.57 V/mm); t is the time; m is the mass of a
single virus (2,484,700 Dalton or 4.1  1018 g as re-
ported by Tito et al. (2000)); Z is the water viscosity
(8.9  104 N s/m2 at 23 1C); and r is the virus radius
(12.5  109 m).
This simple model revealed that under our experi-
mental conditions, the maximum virus electrophoretic
velocity was reached in much less than one second for all
studied pHs. As depicted in Fig. 1, the annular distance
between the anode and the cathode was 1.85 mm and the
hydraulic detention time in the EC unit was 24 s
(excluding recycle). Thus, a virus particle entering at
the maximum radial distance from the anode (1.85 mm)
would have to move towards the anode at a minimum
velocity of 0.077 mm/s in order to reach the anode
before exiting the unit. The electrophoretic velocities
calculated from Eq. (3) were 5.07, 5.33, and 5.37 mm/s at
pHs 6.3, 7.3, and 8.3, respectively. Because the electro-
phoretic velocities were all greater than 0.077 mm/s,
MS2 viruses could readily travel the maximum distance
to the anode during their residence time in the EC unit,
which would result in a higher virus concentration in the
proximity of the anode.
3.10.2. Experimental verification for the improvement of
virus removal at locally high iron and virus concentrations
near the anode
A high-iron, -virus concentration CC-MF process was
designed to simulate the EC-MF process conditions of
locally high iron and MS2 virus concentrations in the
area near the anodes.
The first experiment simulated a local doubling of the
iron and virus concentrations at pH 6.3. During
adsorption and mixing, the doubled iron and virus
concentrations were 2  106 pfu/mL and 10 mg/L, re-
spectively. After adsorption, mixing, and dilution, the
final iron and virus concentrations were 106 pfu/mL and
5 mg-Fe/L, respectively. The experimental procedure
was as follows:
(1) Add source virus into 100 mL source water (DI
water spiked with 1 mM CaCl2, and 3 mM NaH-
CO3). The target virus concentration was
2  106 pfu/mL.
(2) Take samples after 4 s mixing (Sample 1).
(3) Add iron coagulant solution to produce a target
Fe(OH)3(s) concentration of 10 mg/L as Fe.
(4) Take samples after 24 s mixing (Sample 2).
(5) Dilute the mixture back to the desire average 5 mg-
Fe/L and 106 pfu/mL virus concentrations by adding
(6) Mix the diluted sample for 10 s, and take samples for
iron and virus analysis (Sample 3).
(7) Transfer the water to the feed bottle of the MF
system.
(8) Filter the water using MF in the dead-end mode at
constant pressure (69 kPa, 10 psig).
(9) Take samples from the effluent and analyze for virus
concentration (Sample 4).
These experiments were performed at pH 6.3 and in
triplicate as were similar experiments for 4- and 20-times
higher iron and MS2 virus concentrations. The final iron
and virus concentrations (Sample 3) for all of these
experiments were 5 mg-Fe/L, and 106 pfu/mL, respec-
tively. There was no statistically significant difference
between Samples 1 and 2 for any experiment, indicating
that the increased levels of Fe(III) did not affect the
virus viability.
Fig. 11 summarizes the LRVs of all the high-iron,
high-virus concentration experiments based on the
difference between Samples 3 and 4 for each experiment.
Virus removal obtained during the control experiment
was 1.5 log, whereas, greater removals, 2.6-, 3.8-, and 5-
log, were achieved at 2-, 4-, and 20-times higher iron and
virus concentrations, respectively. Thus, locally high
iron and virus concentrations significantly improved
virus removal, which helps explain why EC with its

Log Removal of MS2 Bacteriophage
6.0
5.0
4.0
3.0
2.0
1.0
0.0
0 4 8 12
Iron and Virus Concentration Factor
Fig. 11. Enhancements in virus removal resulting from 2 to 20-
times higher iron and virus concentrations. Final iron
concentration ¼ 5 mg/L, pH ¼ 6.3, final target MS2 virus
concentration ¼ 106 pfu/mL.
locally higher iron and virus concentrations near the
anode was superior to CC for virus removal.
4. Conclusions
Iron electrocoagulation proved to be an excellent
pretreatment for virus removal by MF. The LRV for
MS2 virus suspended in simulated natural water was 4.5
(99.997% removal) when generating 10 mg/L Fe by EC.
As expected, virus removal improved with increasing
iron concentration. Without EC pretreatment, the LRV
for MF was only 0.5 (32% removal).
At low iron dosages and lower pH values (6.3 and
7.3), virus removal by EC-MF was predominantly due
to adsorption of the negatively charged viruses onto the
positively charged iron flocs and the subsequent removal
of the flocs by MF. However, virus removal at high
Fe(III) dosages was attributed to enmeshment prior to
removal by MF. These results suggest that other
clarification technologies capable of removing flocs
(e.g. media filtration) would also be effective for virus
removal, as was MF in this study.
In a comparison of the processes, EC significantly
outperformed chemical coagulation (CC) with ferric
chloride as a pretreatment for MF. For example, at pH
7.3 and an iron dose of 10 mg/L, the LRVs for MS2
virus removal were 4.5 and 2.0 for EC and CC,
respectively. The dramatic improvement in virus re-
moval by EC compared with CC held true for all iron
dosages (2–10 mg/L) and pHs tested (6.3, 7.3, and 8.3).
It was proposed and subsequently verified by modeling
and additional experiments that locally higher iron and
ARTICLE IN PRESS
B. Zhu et al. / Water Research 39 (2005) 3098–3108 3107
virus concentrations within the EC unit improved virus
removal.
These bench-scale results, which demonstrate the
significant advantage of iron EC compared with CC
for virus removal, should be verified at the pilot scale.
Furthermore, the results here suggest that EC might also
be superior to CC for natural organic matter removal
during membrane pretreatment.
Acknowledgments
16 20 We thank Prof. Michael Benedik (Department of
Biology and Biochemistry, University of Houston) and
Dr. Vidya Gopalakrishnan (Department of Molecular
Genetics, The University of Texas, M.D. Anderson
Cancer Center) who instructed us in the methods of
enumerating coliphages. Wendong Xu is acknowledged
for his help in settling up the experiments. This research
was funded by grants from the Texas Higher Education
Coordinating Board (project No. 003652-0412-1999),
the National Science Foundation (BES-0134301), and
the University of Houston. The contents do not
necessarily reflect the views and policies of the sponsors
nor does the mention of trade names or commercial
products constitute endorsement or recommendation for
use.
References
Brock, T.D., Madigan, M.T., 1991. Biology of Microorgan-
isms. Prentice-Hall, Englewood Cliffs, NJ.
Can, O.T., Bayramoglu, M., Kobya, M., 2003. Decolorization
of reactive dye solutions by electrocoagulation using
aluminum electrodes. Ind. Eng. Chem. Res. 42 (14),
3391–3396.
Chen, X., Chen, G., Yue, P.L., 2002. Investigation on the
electrolysis voltage of electrocoagulation. Chem. Eng. Sci.
57 (13), 2449–2455.
Clesceri, L.S., Greenberg, A.E., Eaton, A.D., 1998. Standard
Methods for the Examination of Water and Wastewater,
20th ed. American Public Health Association/American
Water Works Association/Water Environment Federation,
Washington, DC.
Coffey, B.M., Stewart, M.H., Wattier, K.L., Wale, R.T., 1993.
Evaluation of microfiltration for metropolitan’s small
domestic water systems. AWWA Membrane Technology
Conference Proceedings, Baltimore, MD.
Drees, K.P., Abbaszadegan, M., Maier, R.M., 2003. Compara-
tive electrochemical inactivation of bacteria and bacter-
iophage. Water Res. 37 (10), 2291–2300.
EPA, 2001. Low-pressure membrane filtration for pathogen
removal: application, implementation, and regulatory
issues. EPA 815-C-01-001, Office of Water (4601).
Han, M., Song, J., Kwon, A., 2002. Preliminary investigation of
electrocoagulation as a substitute for chemical coagulation.
Water Supply 2 (5-6), 73–76.
 ARTICLE IN PRESS
3108 B. Zhu et al. / Water Research 39 (2005) 3098–3108
Hingston, F.J., 1981. A review of anion adsorption. In:
Anderson, M.A., Rubin, A.J. (Eds.), Adsorption of
Inorganics at Solid–Liquid Interfaces. Ann Arbor Science,
Ann Arbor, MI.
Holt, P.K., Barton, G.W., Wark, M., Mitchell, C.A., 2002. A
quantitative comparison between chemical dosing and
electrocoagulation. Colloids Surf. A: Physicochem. Eng.
Aspects 211, 233–248.
Horner, G., Duffey, J.G., 1983. Electrochemical Removal of
Heavy Metals from Waste Water. The American Electro-
platers Society Annual Meeting, Denver, CO.
Jacangelo, J.G., Laine, J.M., Carns, K.E., Cummings, E.W.,
Mallevialle, J., 1991. Low-pressure membrane filtration for
removing Giardia and microbial indicators. J. Am. Water
Works Assoc. 83 (9), 97–106.
Lahoussine-Turcaud, V., Wiesner, M.R., Bottero, J.Y., Mal-
levialle, J., 1990. Coagulation pretreatment for ultrafiltra-
tion of a surface water. J. Am. Water Works Assoc. 82 (12),
76–81.
Liu, W.K., Brown, M., Elliott, T.S., 1997. Mechanisms of the
bactericidal activity of low amperage electric current (DC).
J. Antimicrob. Chemother. 39 (6), 687–695.
Madaeni, S.S., 1999. The application of membrane technology
for water disinfection. Water Res. 33 (2), 301–308.
Madaeni, S.S., Fane, A.G., Grohmann, G.S., 1995. Virus
removal from water and wastewater using membranes. J.
Memb. Sci. 102 (1-3), 65–75.
Mallevialle, J., Odendaal, P.E., Wiesner, M.R., 1996. Water
Treatment Membrane Processes. McGraw-Hill, New York.
Mills, D., 2000. A new process for electrocoagulation. J. Am.
Water Works Assoc. 92 (6), 34–43.
O’Melia, C.R., 1972. Coagulation and flocculation. In: Weber,
W.J. (Ed.), Physicochemical Processes for Water Quality
Control. Wiley, New York (Chapter 2).
Penrod, S.L., Olson, T.M., Grant, S.B., 1996. Deposition
kinetics of two viruses in packed beds of quartz granular
media. Langmuir 12 (23), 5576–5587.
Roberts, P., 1997. Efficient removal of viruses by a novel
polyvinylidene fluoride membrane filter. J. Virol. Meth. 65
(1), 27–31.
Schindler, P.W., 1981. Surface complexes at oxide-water
interfaces. In: Anderson, M.A., Rubin, A.J. (Eds.), Adsorp-
tion of Inorganics at Solid–Liquid Interfaces. Ann Arbor
Science, Ann Arbor, M.I (Chapter 1).
Tito, M.A., Tars, K., Valegard, K., Hajdu, J., Robinson, C.V.,
2000. Electrospray time-of-flight mass spectrometry of the
intact MS2 virus capsid. J. Am. Chem. Soc. 122 (14),
3550–3551.
Tsouris, C., DePaoli, D.W., Shor, J.T., Hu, M., Ying, T.-Y.,
2001. Electrocoagulation for magnetic seeding of colloidal
particles. Colloids Surf. A: Physicochem. Eng. Aspects 177
(3), 223–233.
Urase, T., Yamamoto, K., Ohgaki, S., 1996. Effect of pore
structure of membranes and module configuration on virus
retention. J. Memb. Sci. 115 (1), 21–29.
Valegård, K., Liljas, L., Fridborg, K., Unge, T., 1990. The
three-dimensional structure of the bacterial virus MS2.
Nature 345, 36–40.
Wiesner, M.R., Veerapaneni, S., Brejchova, D., 1992. Improve-
ments in membrane microfiltration using coagulation
pretreatment. Proceedings of the Fifth Gothenburg Sympo-
sium, Nice, France.
Zhu, B., 2004. The development of an electrocoagulation-
microfiltration process for virus removal from drinking
water. Doctoral Dissertation, University of Houston,
Houston, TX.
Zhu, B., Clifford, D., Chellam, S., 2004. Virus removal by iron
coagulation-microfiltration. Water Research, under revision.