Chemical Engineering Science 57 (2002) 2449 – 2455

www.elsevier.com/locate/ces

Investigation on the electrolysis voltage of electrocoagulation

Xueming Chen, Guohua Chen ∗ , Po Lock Yue
Department of Chemical Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong
Received 29 August 2001; accepted 27 February 2002

Abstract

The relation between electrolysis voltage and the other variables of an electrocoagulation process was analyzed. Theoretical models
describing such a relation were established. Experiments were conducted to con4rm the theoretical analysis and to determine the constants
in the models. Both the theoretical analysis and experiments demonstrated that water pH and 7ow rate had little e8ects on the electrolysis
voltage within a large range. The electrolysis voltage depends primarily on the inter-electrode distance, conductivity, current density and
the electrode surface state. The models obtained can be used to calculate the total required electrolysis voltage for an electrocoagulation
process. ? 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Model; Overpotential; Aluminum electrode; Water; Wastewater

1. Introduction Secondly, it is able to ful4ll simultaneous coagulation and

7otation, with less sludge produced. Thirdly, the electroco-
Electrocoagulation is an e8ective process to destabilize agulation equipment is very compact and thus suitable for
4nely dispersed particles for water and wastewater treat- installation where the available space is rather limited. Fur-
ment. It has been successfully used to treat potable water thermore, the convenience of dosing control only by adjust-
(Vik, Carlson, Eikum, & Gjessing, 1984), urban wastew- ing current makes automation quite easy.
ater (Pouet & Grasmick, 1995) and a variety of industrial Although electrocoagulation has been available for more
wastewaters (Dobolyi, 1978; Pazenko, Khalturina, Kolova, than a century, nowadays the design of an industrial electro-
& Rubailo, 1985; Balmer & Foulds, 1986; Demmin & coagulation unit is still mainly based on empirical knowl-
Uhrich, 1988; Renk, 1988; Do & Chen, 1994; McClung edge due to the lack of available models. The electrolysis
& Lemley, 1994; Lin & Peng, 1996; Chen, Chen, & Yue, voltage is one of the most important variables. It is strongly
2000a, b). dependent on the current density, the conductivity of the
Usually, aluminum or iron plates are used as electrodes water=wastewater to treat, the inter-electrode distance, and
in the electrocoagulation process. When a DC voltage is the surface state of electrodes. A model involving terms
applied, the anodes sacri4ce themselves to produce Al3+ of activation overpotential, concentration overpotential and
or Fe2+ ions. These electrochemically generated metallic ohmic drop of the solution resistance has been proposed
ions are good coagulants. They can hydrolyze near the by Vik et al. (1984). However, that model cannot directly
anodes to produce a series of activated intermediates that predict the electrolysis voltage because it still contains un-
are able to destabilize the 4nely dispersed particles present known terms including the activation overpotential and the
in the water=wastewater to treat. The destabilized particles concentration overpotential. In fact, these unknown overpo-
then aggregate to form 7ocs. At the meantime, the tiny tentials can be related to other simple variables according
hydrogen bubbles produced at the cathode can 7oat most to the Tafel equation and Nernst equation. Therefore, it is
7ocs formed, reaching e8ective separation of particles from possible to obtain simpli4ed models for the estimation of
water=wastewater. Compared with conventional coagula- electrolysis voltage.
tion, electrocoagulation has many advantages. Firstly, it The objectives of the present study are to establish the the-
is more e8ective in destabilizing small colloidal particles. oretical models regarding the electrolysis voltage required
in the electrocoagulation process and to verify them experi-
mentally. Since the dispersed particles in water=wastewater
∗ Corresponding author. Tel.: +852-2358-7138; fax: +852-2358-0054. do not take part in the electrochemical reactions during elec-
E-mail address: [email protected] (Guohua Chen). trocoagulation, the electrochemical behavior performed in

0009-2509/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 9 - 2 5 0 9 ( 0 2 ) 0 0 1 4 7 - 1
2450 X. Chen et al. / Chemical Engineering Science 57 (2002) 2449–2455

real water=wastewater should be similar to that carried out Anode surface

in aqueous solutions. Therefore, in this work, all experi-
ments were conducted in the synthesized Na2 SO4 solutions
for the purpose of easy control of conductivity. In addition,
aluminum electrodes were used because they are the most
common electrodes found in industrial applications.

Al 3+
2. Theoretical analysis

When current passes through an electrochemical reactor,

it must overcome the equilibrium potential di8erence, an-
ode overpotential, cathode overpotential and ohmic poten- H+
tial drop of the solution (Scott, 1995). The anode overpo-
tential includes the activation overpotential and concentra-
_
tion overpotential, as well as the possible passive overpo- OH
tential resulted from the passive 4lm at the anode surface, 0 x
while the cathode overpotential is principally composed of
the activation overpotential and concentration overpotential.
Therefore,
Diffusion layer
d
U0 = Eeq + a; a + a; c + a; p + |c; a | + |c; c | + j: (1)

Fig. 1. Concentration variation of Al3+ , H+ and OH− near the anode.
When aluminum is used as electrode material, there are
three major reactions in the electrochemical reactor as fol-
lows: In an electrochemical reaction, the mass transport includes
(i) The oxidation reaction at the anode, di8usion, convection and electric migration, and can be cal-
culated by Nernst–Plank equation (Bard & Faulkner, 1980)
Al − 3e = Al3+ ;
@Cj (x) Cj (x) @(x)
RT Jj (x) = −Dj − zj FDj + Cj (x)v(x)
’Al3+ =Al = ’oAl3+ =Al
+ ∗
ln CAl 3+ : @x RT @x
3F
or
(ii) The reduction reaction at the cathode,
2H+ + 2e = H2 ; @Cj (x) @(x)
Jj (x) = −Dj + uj Cj (x) + Cj (x)v(x):
@x @x
RT CH∗2+
’H+ =H2 = ’oH+ =H2 + ln : The total current is contributed by all ions present in the
2F PH 2
water=wastewater
(iii) The hydrolysis reaction,
Al3+ + 3H2 O = Al(OH)3 + 3H+ ; j = Fzj Jj


C ∗3+ @Cj (x) @(x)
Kh = ∗H (2) = Fzj −Dj + uj Cj (x) + Cj (x)v(x)
CAl3+ @x @x

The equilibrium potential di8erence between the anode or


 
and the cathode is @Cj (x)
j =  zj F −Dj + Cj (x)v(x) + tj j :
Eeq = ’Al3+ =Al − ’H+ =H2 ; @x
i.e. Near the electrode surface, the convective 7ux term is
RT (pH2 )3=2 eliminated (Scott, 1995). Moreover, within the di8usion
Eeq = ’oAl3+ =Al − ’oH+ =H2 + ln : (3) layer adjacent to the anode surface, except for electrochem-
3F Kh
ically and chemically reactive ions including Al3+ , H+ and
Eq. (3) suggests that Eeq is not a8ected by pH. The acti-
OH− , non-reactive ions such as Na+ , SO2+4 , and so on, can
vation overpotential can be calculated from Tafel equation
produce a gradient of concentration and thus cause a dif-
when the current density is relatively large as is the case for
fusion current which is equal but opposite to the migra-
most industrial installations
tion current at steady state. Consequently, the net trans-
a; a = aa + ba ln j; (4) port and the net current of ions except for Al3+ , H+ , OH−
through the anode di8usion layer is zero. Fig. 1 illustrates
|c; a | = ac + bc ln j: (5) the concentration variation of Al3+ , H+ and OH− near the
 X. Chen et al. / Chemical Engineering Science 57 (2002) 2449–2455 2451

anode. In the practical application, although the original pH Cathode surface

of water=wastewater may be low or high, the in7uent pH is
usually controlled in a mediate range as the best electroco-
agulation eMciency is commonly achieved in that pH range.
Because of this, concentrations of H+ and OH− near the
anode are relatively low and can be neglected. The total cur-
rent in the circuit is then composed primarily of the current
from migration and di8usion of Al3+ , that is
_
@CAl3+ |x=0 OH
j = −3FDAl3+ + tAl3+ j:
@x
Suppose that Al3+ concentration varies linearly across the
whole di8usion layer, that is
∗
@CAl3+ |x=0 CAl3+ |x=0 − CAl 3+
= ; H+
@x a
0 x
then
∗
CAl3+ |x=0 − CAl 3+
j = 3FDAl3+ + tAl3+ j:
a
∗ Diffusion layer
Usually, CAl 3+ CAl3+ |x=0 .

Thus, Fig. 2. Concentration variation of H+ and OH− near the cathode.

(1 − tAl3+ )j a
CAl3+ |x=0 = :
3FDAl3+ from the di8usion and migration of OH−
a depends on the turbulence of water or wastewater 7ow. @COH− |x=0
In the electrocoagulation process, the turbulence is caused j = −FDOH− + tOH− j:
@x
mainly from the agitation of the hydrogen gas generated at
An analogous approximation is to assume a linear varia-
the cathode because the 7ow rate through an electrochemical
tion of OH− concentration across the whole di8usion layer,
reactor is slow. The larger the current density, the more the
which is valid when bulk pH is not far away from neutral.
generated hydrogen gas and thus the thinner the a . Suppose
Then
a power-law relationship exists between a and j, that is
@COH− |x=0 C ∗ − − COH− |x=0
a = k1 j
−p
; = OH :
@x c

then Thus,
1−p ∗
k1 (1 − tAl3+ )j COH − − COH− |x=0
CAl3+ |x=0 = : (6) j = −FDOH− + tOH− j:
3FDAl3+ c
∗ ∗
CAl 3+ can be calculated from Eq. (2) as Usually, COH − COH− |x=0 , thus

CH∗3+ (1 − tOH− )j c
∗
CAl 3+ = : (7) COH− |x=0 = :
Kh FDOH−

Therefore, Similarly, suppose c = k2 j −q , then

RT CAl3+ |x=0 RT k1 Kh (1 − tAl3+ )j 1−p k2 (1 − tOH− )j 1−q
a; c = ln = ln : (8) COH− |x=0 = :
3F ∗
CAl 3+ 3F 3FDAl3+ CH∗3+ FDOH−

Similarly, the net transport and thus the net current of ions The corresponding concentration of H+ at the cathode sur-
except for H+ , OH− through the cathode di8usion layer is face can be calculated according to the ion product of water
zero. Fig. 2 illustrates the concentration variation of H+ and Kw
CH+ |x=0 =
OH− near the cathode. Because H+ is reduced to produce COH− |x=0
hydrogen gas, pH near the cathode is alkaline even if bulk
or
pH is acidic but not extremely strong. In other words, the
concentration of OH− near the cathode is much higher than Kw FDOH−
CH+ |x=0 = : (9)
that of H+ and hence the current there comes predominantly k2 (1 − tOH− )j 1−q
2452 X. Chen et al. / Chemical Engineering Science 57 (2002) 2449–2455

Thus, tOH− , it approaches constant when

is large. This is because
RT CH∗ + tAl3+ and tOH− approach zero at large
. Eqs. (13) and (14)
|c; c | = ln indicate that U0 is independent on pH. The values of A, K1 ,
F CH+ |x=0
K2 , m, n, need to be determined experimentally.
RT k2 (1 − tOH− )j 1−q CH∗ +
= ln : (10)
F Kw FDOH−
Combining Eqs. (3) – (5), (8) and (10), Eq. (1) can be 3. Experimental verication
rewritten as
RT (pH2 )3=2 To con4rm the theoretical analysis, a series of experi-
U0 = ’oAl3+ =Al − ’oH+ =H2 + ln + aa + ac ments was conducted at di8erent pH, 7ow rate, current den-
3F Kh
sity, conductivity and anode surface state. The experimental
RT k1 Kh (1 − tAl3+ ) setup is schematically shown in Fig. 3. It consists of an elec-
+ ln trocoagulation system, a DC power supply (PD 110 −5 AD,
3F 3FDAl3+
Kenwood TMI Corporation, Japan), a feed tank and a micro-
RT k2 (1 − tOH− ) d processor pump (Model 7518-12, Master7ex, Cole-Parmer
+ ln + a; p + j Instrument Co., USA). The electrocoagulation system has
F Kw FDOH−

an electrochemical reactor of 0:30 L and a separator of 1:2 l.


RT (1 − p) RT (1 − q) The electrochemical reactor contains 4ve aluminum elec-
+ ba + bc + + ln j: (11) trodes connected in a bipolar mode. The original dimension
3F F
of each electrode is 140 mm × 44 mm × 3 mm. Water 7ows
Let
through the electrochemical reactor upward, perpendicular
RT (pH2 )3=2 to the electric current. Deionized water was used in all the
A = ’oAl3+ =Al − ’oH+ =H2 + ln + aa + ac
3F Kh experimental runs. The conductivity and pH were adjusted
by adding Na2 SO4 (100 g=L) solution and H2 SO4 (0:1 M)
RT k1 Kh (1 − tAl3+ ) RT k2 (1 − tOH− )
+ ln + ln or NaOH (0:1 M) solution. Water pH and conductivity were
3F 3FDAl3+ F Kw FDOH− measured using pH meter (Model 420A, Orion Research Inc,
and USA) and conductivity meter (Checkmate 90, Corning In-
RT (1 − p) RT (1 − q) corporated Scienti4c Products Division, USA), respectively.
K1 = b a + b c + + ;
3F F
then Eq. (11) becomes
d
U0 = A + a; p + j + K1 ln j: (12)

6 _
It should be noted that the passive overpotential highly +
depends on the electrode surface state. For the new
non-passivated aluminum electrodes, the passive overpo-
tential can be neglected and Eq. (12) simpli4es to Sludge
d Effluent
U0 = A + j + K1 ln j: (13)

For the old passivated aluminum electrodes, the passive

overpotential is usually signi4cant. It is related to many fac-
tors including pH, conductivity and current density. How-
ever, taking into account the fact that pH close to the anode
is always acidic as long as bulk pH is not overly alkaline,
it can be generally believed that a; p depends mainly on 5
the conductivity and current density. Usually, a; p increases
with the current density and decreases with the conductivity. 3
Assuming a power-law relation of a; p with j and
, then 4
K2 j n
a; p = m :

Influent
2
Therefore, for old passivated aluminum electrodes 1
d K2 j n
U0 = A + j + K1 ln j + m : (14) 1. Tank 2. Microprocessor Pump 3. Electrocoagulation System


4. Electrochemical Reactor 5. Separator 6. D.C. Power Supply
On the right-hand side of Eqs. (13) and (14), both K1
and K2 are constants. Although A is related to tAl3+ and Fig. 3. Schematic diagram of experimental setup.
 X. Chen et al. / Chemical Engineering Science 57 (2002) 2449–2455 2453

The experiments were carried out in two stages. The 4rst 25

stage experiments were conducted after the electrodes had κ = 355 µs/cm

been used for about a month. At this stage, the anode surface 20 κ = 578 µs/cm
looked still relatively smooth, thus could be considered as
κ = 908 µs/cm
not passivated. The second stage experiments were done
κ = 1210 µs/cm
after the electrodes had been used for more than 3 months. 15

U0 /V
At this stage, the anode surface was full of pits and covered κ = 2090 µs/cm

with metal oxide, i.e. the anodes have been passivated. 10

4. Results and discussion

0
4.1. E.ect of pH 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Current density/A m−2

The theoretical analysis has demonstrated that the elec- Fig. 4. Dependence of electrolysis voltage on conductivity and current
trolysis voltage between electrodes is independent of pH as density for non-passivated electrodes. Inter-electrode distance 6:4 mm, pH
long as water is not far away from neutral. In order to con- ◦
7.20 –7.37, temperature 22.5 –23:8 C, 7ow rate 6 l=h.
4rm it, electrolysis voltages were measured at di8erent pH
values. It was found that the e8ect of pH on U0 was really
insigni4cant for both new and passivated electrodes. For the
new electrodes, an increase in pH from 3.75 to 10.41 re- 30
sulted in only an increase in U0 from 13.2 to 13:8 V even κ = 445 µs/cm
at a current density as high as 137 A=m2 . For the passivated 25 κ = 849 µs/cm
electrodes, when pH varied from 3.38 to 10.79, the maxi- κ = 1390 µs/cm
mum di8erence of U0 measured at a constant current den- 20 κ = 2120 µs/cm
sity was only 7.7%. Therefore, both results from new and κ = 3130 µs/cm
passivated electrodes support the theoretical analysis well.
U0 /V

15

4.2. E.ect of water /ow rate 10

In the theoretical analysis, the water 7ow rate was as- 5

sumed not to a8ect the electrolysis voltage signi4cantly.

This needs to be veri4ed experimentally. Although a slight 0
0 10 20 30 40 50 60 70 80 90 100
decrease in U0 was found as 7ow rate increased from 3 L=h
Current density/A m−2
to 15 L=h using new electrodes, the extent of such a de-
crease was always less than 10%. In contrast, when the cur- Fig. 5. Dependence of electrolysis voltage on conductivity and current
rent density varied from 20 to 100 A=m2 , almost 4vefold in- density for passivated electrodes. Inter-electrode distance 7:0 mm, pH
◦
crease in electrolysis voltage was measured. For passivated 6.95, temperature 21.9 –22:8 C, 7ow rate 6 l=h.
electrodes, it was found that the electrolysis voltage was al-
most the same within the investigated water 7ow rate range
of 1.8–21:6 L=h. This phenomenon is expected because the
rough surface of passivated electrodes makes the e8ect of It shows in Fig. 4 that good agreement exists between
7ow rate on turbulence insigni4cant. Such results support the measured and predicted U0 for large
solution. This is
the assumption made earlier regarding the e8ect of 7ow rate. expected because A is a constant only when
is large. Al-
though there is considerable di8erence in
for di8erent in-
4.3. Dependence of U0 on j and
dustrial operation, it is usually within the range investigated
in this study. Thus, the present model can be applied without
Fig. 4 demonstrates the electrolysis voltage between elec- much error. The constant A has a negative value, which is
trodes as a function of the conductivity and current density attributed to the large but negative standard electrode poten-
for non-passivated electrodes. Through nonlinear regression tial di8erence between the anode and the cathode, −1:706 V
◦

of the data in Fig. 4, the constant and coeMcient in Eq. (13) at 25 C.

were obtained as A = −0:76, K1 = 0:20. Hence Better prediction was obtained for passivated electrodes
as shown in Fig. 5. This may be because the power-law
d term with regard to j accommodates partially the variation
U0 = −0:76 + j + 0:20 ln j: (15)

of A with
. Again the constants in Eq. (14) were obtained
2454 X. Chen et al. / Chemical Engineering Science 57 (2002) 2449–2455

through nonlinear regression, and the equation became p H2 fractional pressure of hydrogen at the cathode, atm
0:016 d R gas constant, J=mol K
U0 = −0:43 + 0:47 j 0:75 + j + 0:20 ln j: (16) T absolute temperature, K

tj transport number of species j

With U0 obtained, the total required electrolysis voltage U
U total required electrolysis voltage of an electroco-
of an electrocoagulation process can be calculated easily. In
agulation process, V
general, there are two basic electrode connection modes: the
uj mobility of species j, m2 =V s
monopolar mode and the bipolar mode. For the monopolar
U0 electrolysis voltage between electrodes, V
mode, the total required electrolysis voltage is the same as
v(x) convective velocity of water 7ow in the current
the electrolysis voltage between electrodes, that is
direction at distance x, m=s
U = U0 : (17) zj charge number of species j

For the bipolar mode, the total required electrolysis volt- Greek letters
age is U0 times the number of total cell which is the number
of electrodes minus one. Thus a di8usion layer thickness near an anode, m
c di8usion layer thickness near an cathode, m
U = (N − 1)U0 : (18) a; a anode activation overpotential, V
a; c anode concentration overpotential, V
a; p anode passive overpotential, V
5. Conclusions c; a cathode activation overpotential, V
c; c cathode concentration overpotential, V
Theoretical analysis and experiments demonstrated that
conductivity of water=wastewater treated, mho=m
water pH and 7ow rate had little e8ects on the electroly- (x) potential at distance x, V
sis voltage of electrocoagulation process. Two mathematical
models, one applicable to non-passivated aluminum elec-
trodes and the other to passivated aluminum electrodes were Acknowledgements
established and veri4ed. With these models the total required
electrolysis voltage of an electrocoagulation process can be The authors wish to acknowledge the Environment and
calculated. Conservation Fund=WooWheelock Green Fund for the 4-
nancial support of this project.

Notation References

Balmer, L. M., & Foulds, A. W. (1986). Separation oil from oil-in-water

aa constant of Tafel equation at the anode, V emulsions by electro7occulation=electro7otation. Filtration and
ac constant of Tafel equation at the cathode, V Separation, 23(11=12), 366–369.
ba coeMcient of Tafel equation at the anode, V Bard, A. J., & Faulkner, L. R. (1980). Electrochemical methods. New
bc coeMcient of Tafel equation at the cathode, V York: Wiley.
Chen, G. H., Chen, X. M., & Yue, P. L. (2000a). Electrocoagulation and
Cj∗ bulk concentration of species j, mol=l electro7otation of restaurant wastewater. Journal of Environmental
Cj (x) concentration of species j at distance x, mol=l Engineering, 126, 858–863.
Cj |x=0 concentration of species j at the electrode surface, Chen, X. M., Chen, G. H., & Yue, P. L. (2000b). Separation of pollutants
mol=l from restaurant wastewater by electrocoagulation. Separation and
d net distance between electrodes, m Puri8cation Technology, 19, 65–76.
Demmin, T. R., & Uhrich, K. D. (1988). Improving carpet wastewater
Dj di8usion coeMcient of species j, m2 =s treatment. American Dyestu. Reports, 77, 13–18, 32.
Eeq equilibrium potential di8erence between an anode Do, J. S., & Chen, M. L. (1994). Decolorization of dye-containing
and a cathode, V solutions by electrocoagulation. Journal of Applied Electrochemistry,
F Faraday constant, C=mol 24, 785–790.
j current density, A=m2 Dobolyi, E. (1978). Experiments aimed at the removal of phosphate by
electrochemical methods. Water Research, 12, 1113–1116.
Jj (x) 7ux of species j at distance x, mol=m2 s Lin, S. H., & Peng, C. F. (1996). Continuous treatment of textile
Kh hydrolysis constant wastewater by combined coagulation, electrochemical oxidation and
Kw ion product of water activated sludge. Water Research, 30, 587–592.
m constant McClung, S. M., & Lemley, A. T. (1994). Electrochemical treatment
and HPLC analysis of wastewater containing acid dyes. Wastewater
n constant
Treatment, 26, 17–22; methods. Water Research, 12, 1113–1116.
N total electrode number of an electrocoagulation unit Pazenko, T. Ya., Khalturina, T. I., Kolova, A. F., & Rubailo, I. S. (1985).
p constant Electrocoagulation treatment of oil-containing wastewaters. Journal of
q constant Applied USSR, 58, 2383–2387.
 X. Chen et al. / Chemical Engineering Science 57 (2002) 2449–2455 2455

Pouet, M. F., & Grasmick, A. (1995). Urban wastewater treatment by Scott, K. (1995). Electrochemical processes for clean technology (pp.
electrocoagulation and 7otation. Water Science and Technology, 31, 12– 62). Cambridge, UK: The Royal Society of Chemistry.
275–283. Vik, E. A., Carlson, D. A., Eikum, A. S., & Gjessing, E. T. (1984).
Renk, R. R. (1988). Electrocoagulation of tar sand and oil shale Electrocoagulation of potable water. Water Research, 18, 1355–1360.
wastewaters. Energy Progress, 8, 205–208.