Catalysis Today 193 (2012) 120–127

Contents lists available at SciVerse ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

Experiment and modeling of advanced ozone membrane reactor for treatment of

organic endocrine disrupting pollutants in water
Hung Lai Ho a , Wai Kit Chan a , Angelique Blondy a , King Lun Yeung a,b,∗ , Jean-Christophe Schrotter c
a
Department of Chemical and Biomolecular Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong
b
Division of Environment, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong
c
Water Research Center of Veolia, Anjou Recherche, Chemin de la Digue, BP 76. 78603, Maisons Laffitte, Cedex, France

a r t i c l e i n f o a b s t r a c t

Article history: An advanced ozone membrane reactor that uses membranes for ozone distribution, reaction contact and
Received 19 October 2011 selective water separation was used for ozone treatment of a recalcitrant endocrine disrupting compound
Received in revised form 15 February 2012 in water. Experiments and model calculation were employed to examine the ozonation of phthalate
Accepted 28 March 2012
in the new reactor. Experimental results showed that fast ozone mass transfer rate is responsible for
Available online 12 May 2012
membrane reactor’s superb performance compared with a semibatch reactor. Selective water removal
further enhanced phthalate conversion and TOC removal by concentrating the pollutants in the reaction
Keywords:
zone. Clean water was produced by membrane separation. Mathematical model was used to investigate
Water and wastewater treatment
Ozonation reaction
the effect of membranes, reactor design and reaction operation on pollutant conversion and removal.
Phthalate © 2012 Elsevier B.V. All rights reserved.
Nanofiltration membrane
Zeolite membrane

1. Introduction Ozone and membrane processes are technologies that showed

There is growing evidence that a large number of chemical
compounds found in common household products ranging from
medicines, cosmetics and personal care products, and household
cleansers, can survive state-of-the-art wastewater treatment pro-
cesses to contaminate surface and ground waters [1–10]. Many
of these compounds are endocrine disruptors and studies carried
out in 2001 [11] and 2003 [12] showed that a large numbers of
endocrine disrupting chemicals (EDCs) survive conventional drink-
ing water treatment processes to persist in finished, potable water.
Several of the compounds were even found in samples of human
blood, milk and urine [13]. Although health risk from chronic expo-
sure to EDCs in humans has not been adequately addressed, their
effects on normal hormonal processes is well documented and
there is extensive evidence of their adverse effects in wildlife [14].
It is particularly disturbing that most studies in animal models [15]
showed that early life stages are the most vulnerable to the actions
of EDCs, putting pregnant women and children at greater risk
[16].
∗ Corresponding author at: Department of Chemical and Biomolecular Engineer-
ing, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon,
Hong Kong. Tel.: +852 23587123; fax: +852 23580054.
E-mail address: kekyeung@ust.hk (K.L. Yeung).
the best promise for treatment of EDCs in water [17,18]. How-
ever, ozone treatment alone suffered from slow mineralization
rates [19] and the UV/O3 and O3 /H2 O2 used by various authors
[20–22] to remedy this shortcoming are often expensive and com-
plex. Nanofiltration and reverse osmosis membrane can remove
most EDCs [23,24], but the separated EDCs require further treat-
ment. In addition, membrane fouling is a concern [25]. Ozone has
been used in membrane to alleviate membrane fouling by organic
matters [26,27] and improves membrane filtration processes [28].
Shanbhag et al. [29] explored in their early work the use of a sil-
icone capillary membrane for ozone distributor for treatment of
chemical micropollutants in water. More recently Karnik et al.
[30] employed a catalytic membrane based on ultrafiltration and
ozonation for drinking water treatment. This work examines an
advanced ozone membrane reactor for treatment of recalcitrant
organic EDCs in water. The membrane reactor uses membranes
for ozone distribution, reaction contactor and water separation
in a compact unit that synergistically combines ozone oxidation
and membrane separation to achieve greater treatment efficiency.
The use of multi-functional membrane reactor was shown to
benefit gas-phase and liquid phase reactions [31–38]. Indeed,
prior works showed the new membrane reactor increased min-
eralization rate, improved ozone utilization, reduced membrane
fouling and enhanced clean water production [39,40]. This study
investigates the design of the reactor and membranes using a math-
ematical model to guide the optimization and scale-up the process

0920-5861/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cattod.2012.03.059
 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127 121

Nomenclature
j the jth alumina support sub-model
Ac cross-sectional area (m2 ) p permeate
Ag surface area for gas distribution (m2 ) r retentate
Am membrane surface area (m2 ) Al alumina support phase
C concentration (mol m−3 ) G bulk gas phase
Co feed concentration (mol m−3 ) L bulk liquid phase
Cv convective flux (mol m−2 s−1 ) Sat saturated vapor pressure (bar)
D diffusivity (m2 s−1 )
De depletion factor
Di diffusive flux (mol m−2 s−1 ) for practical application in the treatment of EDCs-contaminated
En enhancement factor water.
Ni organic compounds permeation flux (mol m−2 s−1 )
No total number of species 2. Experiment and model
Nw water permeation flux (m3 m−2 s−1 )
P water permeance (m2 bar−1 s−1 ) 2.1. Potassium hydrogen phthalate (KHP) ozonation reaction
P EDC permeance (m−1 s−1 ) kinetics
Pe Peclet number
Q̇ gas feed rate (m3 s−1 ) The KHP ozonation reaction scheme in Fig. 1 was mainly based
R reaction rate (mol m−3 s−1 ) on the experimental results [39], with some of the pathways
Rm membrane intrinsic resistance (bar s m−1 ) deduced from literature reports on dimethyl phthalate (DMP) [41].
Rcp concentration polarization (bar s m−1 ) The main intermediates and products include large intermediates
X dimensionless radial location in the membrane sup- with molecular structure similar to KHP (I1), intermediate ketones,
port aldehydes and carboxylic acids (I2) and simple carboxylic acids
K ozonation reaction rate constant (m3 /mol s) such as formic and acetic acids (I3) that have low UV detection. The
ab specific interfacial area of gas bubble (m2 m−3 ) reaction rate equations are fitted to a second order reaction kinetic
kDd ozone direct decomposition rate constant (s−1 ) typical for ozonation reactions [42,43]. The reaction rates were
kDi Ozone indirect decomposition rate constant obtained from a semibatch reactor using a glass sparger for ozone
(mmol−0.5 s−1 ) distribution. Ozone was produced from pure oxygen by corona dis-
kH Henry’s constant of ozone charge in Wedeco ozone generator. The gas pressure and ozone
kL mass transfer coefficient of ozone from gas to water concentration were measured by an on-line pressure gauge and
(m s−1 ) ozone analyzer (BMT 964® ), respectively. The ozone gas volumet-
p Vapor pressure (bar) ric flowrate was regulated by a Teflon flow meter (Cole Parmer® ).
q̇ gas feed flux through s.s. membrane ozone distriutor The 250 mL semibatch reactor was filled with 150 mL of 250 ppmC
(m s−1 ) KHP solution and sparged with 200 sccm 120 ppm O3 /O2 . Samples
r radius (m) were taken at fixed time intervals and purged with nitrogen gas.
rin inner radius of membrane support (m) 10 mL samples were poured into a 50 mL conical flask and the dis-
rout outer radius of membrane support(m) solved ozone was measured by iodometric titration. 0.3 g KI and six
r transformed radius (m) drops of 0.4 M sulfuric acid were added to the solution and titrated
t time (s) using 0.002 M Na2 S2 O3 solution and starch indicator. The pH was
v superficial flow velocity (m s−1 ) measured and the organics were analyzed by Water Acquity UPLC
v̄ superficial flow velocity per unit reactor height (s−1 ) equipped with BEH C18 column and a PDA e␭ UV–vis detector and
x mole fraction Shimadzu total organic analyzer (TOC-V CSH). The concentrations
y dimensionless concentration of the intermediates and products were obtained by calculating
z stoichiometric coefficient their sensitivity factors (SFi ) from Eqs. (1) and (2).

Greek symbol 
N

˛ fouling factor (m2 s bar mol−1 ) TOCCal = SFi xPAi (1)

ˇ gas volume holdup i=1
ıAl alumina support thickness (m)
ım membrane thickness (m) 
Nrt

εAl porosity SSETOC = (TOCMea − TOCCal )2 (2)

 dimensionless time i=1

where PAi is the peak area of species i (AU); SFi is the sensitiv-
Subscripts ity factor for species i (ppmC/AU) obtained at minimum SSETOC
i different species (i.e., squared errors of TOC); TOCCal is the calculated TOC (ppmC);
ref reference TOCMea is the measured TOC (ppmC); and N is number of species.
w water
A ozone
2.2. Advanced ozone membrane reactor experiments
B endocrine disrupting compound
BP degradation byproducts
A schematic drawing of the advanced ozone membrane reactor
is shown in Fig. 2. Ozone gas bubbles were fed by a 35 mm long
Superscripts
porous stainless steel membrane (i.e., 0.2 ␮m) welded to the stain-
g gas
less steel reactor shell. The stainless steel membrane was purchased
from Mott Metallurgical and had inner and outer diameters of 12.5
122 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127

HO
+ -
K O KHP

O O O
OH
HO CHO HO CHO HO
OOH
+ - + - + -
K O CHO K O HC K O
OH
O O O
O O
OH COOH
HO HO
+ - + -
K O K O
OH CHO
O Intermediate group 1 (I1) O

O OOH O

HO CHO HO CHO HO COOH

COOH CHO CHO CHO HCOOH

Intermediate group 2 (I2) and Low UV-VIS absorptivity intermediates (I3)

CO2 and H2O

Fig. 1. Proposed mechanism for ozonation of KHP molecules in water.

and 15.5 mm, respectively. The membrane was heat treated in air
at 773 K overnight, cleaned in 0.1 N HCl solution, before rinsing
with water and ethanol. This pretreatment ensures fine gas bub-
bles were generated uniformly over the entire membrane. Alumina
membrane from Pall-Exekia with inner and outer diameters of 6.5
and 10 mm, were cut into 75 mm lengths and end-sealed with glass
enamel. A thin layer of zeolite was grown on the inner tube surface
for water separation, while the coarse ␣-Al2 O3 ceramic tube served
as the membrane contactor. The zeolite membrane was grown by
seeding and regrowth method, but unlike previous works [44–50], a
template-free synthesis solution was used to grow the ZSM-5. Seed-
ing is important for controlling the zeolite growth and membrane
properties [51–55]. The membrane separator/contactor was held
in place by a pair of O-rings, giving the advanced ozone membrane
reactor a reactor volume of 2 cm3 , membrane distributor area of
13.7 cm2 and membrane separator/contactor area of 9.2 cm2 . Thus,
the membrane area-to-reactor volume ratio for the membrane dis-
tributor was 690 m−1 , and 460 m−1 for the membrane separator.
The advanced ozone membrane reactor operates in continuous
mode. A 250 ppmC phthalate solution was prepared from potas-
sium hydrogen phthalate (KHP, 99.9%, Sigma–Aldrich). The solution
was fed to the membrane reactor by a Watson-Marlow peristaltic
pump. Once steady-state flow condition was reached, 130 ppm
O3 /O2 gas mixture was bubbled through the membrane distributor
at a flow rate of 20 sccm. The ozone gas mixture was produced by
Wedeco GSO 20 ozone generator and the ozone concentration was
monitored by a set of internal and external (BMT) ozone analyzers.
The ozone gas flow was regulated by Teflon flow meter from Cole-
Palmer. The gas and liquid from the retentate outlet were separated,
and the disengaged ozone gas was analyzed to measure the outlet
concentration. Samples of the retentate liquid were titrated for dis-
solved ozone, and the remaining liquid was purged with nitrogen to
remove the dissolved ozone and quench the reaction. The organ-
ics in the retentate liquid were analyzed by Water Acquity UPLC
and Shimadzu TOC analyzer. Reactions carried out using the mem-
brane distributor and contactor will be referred to as membrane
ozonication.
Water separation and removal was accomplished by pervapo-
ration across the zeolite membrane deposited on the inner surface
of the membrane separator/contactor unit. A vacuum pressure of
60–100 Pa was maintained in the permeate-side by BOC-Edward
vacuum pump. Water was selectively pervaporated across the
zeolite membrane, while the organics were retained in the reac-
tion zone for deeper oxidation. The clean permeate stream from
 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127 123

Fig. 2. Schematic drawing of the advanced ozone membrane reactor.

the advanced ozone membrane reactor was collected in a cooled permeation (Eq. (19)), convective (Eq. (20)) and diffusive fluxes
condenser and samples were withdrawn regularly. The amount (Eq. (21)). Based on the concentrations at the boundary between
was weighed to determine the average flux. The composition and the membrane separator and the permeate section, the permeate
organic carbon content of the retentate and permeate samples were fluxes of water and organics were calculated and their concentra-
analyzed by UPLC and TOC to calculated the overall KHP and TOC tions in the permeate stream can be obtained (cf. Eqs. (16) and
removal efficiencies. (20)).
Material balance equations for gas phase (for hO3 f ≥ h ≥ hO3 o ):
2.3. Advanced ozone membrane reactor modeling ozone gas
G − C G )V̄ g A − k a ˇ((C G /k ) − C L )
dCAG (CoA C L b H
The advanced ozone membrane reactor consisted of three sec- = A A A
for h ≥ hO3 o (3)
tions; the retentate, membrane and permeate. The reactor was dh Vg
considered to be isothermal and isobaric. A uniform ozone distri- gas phase boundary conditions:
bution by the membrane distributor was assumed in the model.
The fluid flows in the retentate and membrane were considered at h = hO3 o : CAG = CoA
G
(4)
along the axial and radial directions, and the diffusions of the
organics in the membrane were calculated using a transient 1D Material balance equations for liquid phase (for hf ≥ h ≥ 0): dis-
convection and diffusion model. A pseudo steady-state, plug flow solved ozone:
condition was assumed for the retentate stream and well-mixed j
dCAL kL ab ˇ((CAG /kH ) − CAL ) + RAL + DifA
for the permeate stream. The retentate stream and membrane sep- = (5)
arator/contactor were coupled by convective and diffusive fluxes dh Vr
of ozone, reactants and products across their boundaries. Water KHP and reaction products:
and organic flux across the ZSM-5 membrane was modeled using
j
experimental and diffusion data, and assumed to be uniform. Thus, dCiL RiL + Difi
the composition of the permeate stream can be calculated from the = (6)
dh Vr
component fluxes across the zeolite membrane. Ozone flux across
the membrane was assumed negligible, which is consistent with Liquid phase boundary conditions:
experimental observation. Thus, ozone oxidation reaction can be
ignored in the permeate stream. It was further assumed that the at h = 0 : CAL = O, CBL = CoB
L
, CjL = 0 (7)
membrane was inert and do not catalyze the reaction.
Material balance equations in the porous membrane contactor
The material balance in the retentate section is shown in Eqs.
For ozone, KHP and its degradation intermediates:
(1)–(5) for ozone gas (Eqs. (3) and (4)), dissolve ozone in liquid
(Eqs. (5) and (7)) and organic compounds (Eqs. (6) and (7)). The
 Al,j

Al,j Al,j ∂2 Cj Al,j
∂Ci j ∂Ci 1 ∂Ci Al,j
concentrations of ozone and organics within the porous membrane + Nw − DiAl + − Ri =0 (8)
contactor are accounted for by the component material balance ∂t ∂r ∂r 2 r ∂r
Eqs. (8)–(11), and their concentration profiles were calculated by
finite element orthogonal collocation method. These equations initial condition:
along with the superficial gas velocity (Eqs. (12) and (13)) and Al,j
retentate liquid velocity (Eqs. (14) and (15)) were solved using the at t = 0 : Ci = 0 for rin ≤ r ≤ rout (9)
124 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127

boundary conditions:at r = rout : Table 1

 Reaction rate equations.
Al,j 
∂C
DiAl i 
 j Al,j L,j Species Rate equations
− Nw (Ci |r=rout − Ci ) = 0 (10)
∂r  dC L
r=rout Ozone RA = A
dt
= −zb2 kb CAL CBL − zc2 kc CAL CCL − zd2 kd CAL CDL − zu2 ku CAL CuL −
3/2
kDd CAL − kDi CAL
and at r = rin : dC L
 KHP RB = B
= −kb CAL CBL

Al,j
dt
dC L
∂Ci  j Al,j j I1 RC = C
= zc1 kb CAL CBL − kc CAL CCL
DiAl

∂r 
− Nw (Ci |r=rin + ni ) = 0 (11) dt
dC L
I2 RD = dt
D
= zd1 kb CAL CBL + zd3 kc CAL CCL − kd CAL CDL
r=rin dC L
I3 RU = dt
U
= zu3 kc CAL CCL + zu4 kd CAL CDL − ku CAL CUL
The superficial gas flow velocity (q̇) is calculated from:
dvg
= 0 for h < hOε o and h > hOε f (12) This allows the reaction to be modeled from five reaction rate
dh
equations (cf. Table 1), instead of eighteen if all reaction intermedi-
dvg q̇Ag ates and products are considered. The organic molecules can react
= = V̄ g for hO3 o ≤ h ≤ hO3 f (13)
dh LOε t Ac directly with the dissolve ozone, or with the hydroxyl radicals gen-
While the retentate liquid flow velocity (v̇r ) is obtained from: erated when ozone decomposes. The latter reaction was ignored
as it is negligible at room temperature and solution pH less than 5
dvr [56]. According to Hoigne et al. [42,43] ozone reactions with most
= 0 for h > hmf (14)
dh organic compounds can be adequately described by an irreversible,
dvr N Am
j second order reaction rate equation. Thus, the ozonation of KHP and
= w for 0 ≤ h ≤ hmf (15) organic intermediates and products (I1, I2 and I3) were described
dh Lmt Ac
by second order rate equations shown in Table 1.
Also, Fig. 3a plots the semibatch ozonation of KHP at room tempera-
Permeate concentration (CiP ): ture. The concentrations of KHP, I1, I2 and I3 are plotted in Fig. 3a
NoAl j along with the total organic carbon (TOC) content. The results show
ni
j=1 that in a semibatch reactor, KHP was completely reacted in 15 min
CiP = NoAl (16)
j of ozonation to produce the reaction intermediates I1. The I1 inter-
j=1
Nw
mediates are more reactive than KHP (cf. Table 2) and react rapidly
Gas volume holdup (ˇ):

ˇ = 8.42(vg ) (17) 300

Lumped ozone mass transfer coefficient (kL ab ): a

Concentraon(mg/L C)

250
−0.246
kL ab = 0.1995(q̇) (18)
200
j
Permeation flux (Nw ):
150
 j j p
−Nw = Pw (pSat
w xw |r=rin − pw ) (19)
100
j
Solute permeation flux (Ni ):
50
−Ni = Pi (Ci
j Al,j p
|r=rin − Ci ) (20)
0
j 0 10 20 30 40 50 60 70
Convective flux (Cvi ):
Time(min)
j j L,j
Cvi = Nw Ci (21) KHP I1 I2 I3 TOC
j
Diffusive flux (Difi ): 300
Calculated concentraon (mg/L C)

 
Al,j  b
j ∂Ck  250
Difi = −DiAl  (22)
∂r 
r=rout 200

3. Results and discussion 150

3.1. KHP ozonation in semibatch reactor 100

The simplified reaction mechanism between phthalate and 50

ozone in Fig. 1, organizes the reaction products into three main
groups (i.e., I1, I2 and I3). It is believed that ozone reacts with KHP 0
according to Criegee mechanism where molecular ozone forms pri- 0 50 100 150 200 250 300
mary ozonide with the unsaturated bonds in the aromatic ring Experimental concentraon(ppmC)
before decomposition into carbonyl and carboxylic compounds
Fig. 3. (a) Plots of KHP, intermediates, products and TOC during ozonation of
shown in Fig. 1. I1 consists of large intermediates deduced from
250 ppmC KHP solution at 25 ◦ C in a semibatch reactor. (b) A plot of calculated and
the previous study on DMP ozonation. I2 are products of ozone experimental concentrations of KHP and organic carbons from semibatch reactions
reaction with I1 intermediates and includes citric, adipic and mal- at 25, 40 and 60 ◦ C. Please note that symbols – experimental data and lines – model
onic acids, while I3 are mainly simple (C1–C4) carboxylic acids. calculation ([O3 ] = 120 ppm, QO3 = 200 sccm, and Vrxtr = 150 mL).
 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127 125

Table 2 100%
Reaction rate constant, activation energy and stochiometry from semibatch
ozonation. a
80%

KHP removal efficiency

Species Parameter Value at different temperature Unit

25 40 60
60%
KHP Ab 3.87 m3 /(mol s)
Eab 12036 J/mol
zb 1 0.5 0.2 – 40%

I1 Ac 0.218 m3 /(mol s)
Eac 3111 J/mol 20%
zc1 6 –
zc2 0.05 0.03 0.02 –
0%
I2 Ad 0.263 m3 /(mol s)
0 2 4 6 8 10 12
Ead 11085 J/mol
Retenon me(min)
zd1 0.7 –
zd2 0.3 0.25 0.08 –
zd3 0.1
100%
m3 /(mol.s)
I3 Au
Eau
677
33263 J/mol
b
80%

TOC removal efficiency

zu2 0.4 0.30 0.08 –
zu3 0.7 –
zu4 1 –
60%

40%
with ozone to produce less toxic I2 and I3 products. The I3 prod-
ucts consisting of simple carboxylic acids are refractory to ozone 20%
(cf. Table 2) and are responsible for the high TOC content of the
reaction mixture (Fig. 3a). They also have larger stoichiometric
0%
coefficient and consume more ozone per molecule than KHP, I1 and 0 2 4 6 8 10 12
I2. Fig. 3a shows that there is excellent agreement between exper-
Retenon me(min)
imental data and model calculations. A semibatch reactor model
based on the work of Benbelkacem and De bellefontaine [57] was Fig. 4. Percent removals of (a) KHP and (b) TOC in the advanced ozone membrane
used in the calculation along with the experimental rate kinetics in reactor with (×, +) and without membrane pervaporation of water (, ). Reactions
were carried out at 25 ◦ C (, ×) and 40 ◦ C (, +). Please note that symbols – exper-
Tables 1 and 2. The calculated and experimental concentrations
imental data and lines – model calculation ([KHP] = 250 ppmC, [O3 ] = 130 ppm, and
of KHP and organic carbon from experiments carried out at 25, QO3 = 20 sccm).
40 and 60 ◦ C were plotted in Fig. 3b. The plot shows that there is
good agreement between experiment and model over the different
temperatures and reaction conditions used in the study.
mass transfer rate from the fine gas bubbles produced by the
stainless steel membrane distributor. The pressure drop across the
3.2. KHP ozonation in advanced ozone membrane reactor membrane distributor is also small and it requires less energy to
generate fine bubbles. Mathematical calculations were done to
KHP ozonation was performed in the advanced ozone mem- optimize ozone usage in the advanced ozone membrane reactor.
brane reactor shown in Fig. 2. The reaction was carried out with and Fig. 5 plots the calculated KHP and TOC removal as a function of
without water separation across the zeolite pervaporation mem- ozone dosage (i.e., moles of ozone per mole C) for different volumet-
brane. The reaction at 25 and 40 ◦ C are plotted in Fig. 4 as a function ric feed flow rates of ozone. The calculations show that a complete
of hydraulic retention time. The data show that a retention time of KHP conversion can be obtained using ozone dosage as little as 2,
6 min in the membrane reactor is sufficient to reacts all KHP in
the water (Fig. 4a) and temperature affects TOC removal (Fig. 4b)
more than KHP conversion during membrane ozonation. Selec-
tive water separation by zeolite membrane pervaporation during
advanced ozone membrane reactor operation resulted in a signifi-
cant enhancement in KHP and TOC removals. KHP was completely
reacted within 3 min of ozonation (Fig. 4a) and TOC removal was
doubled compared to membrane ozonation (Fig. 4b). Model calcu-
lations were conducted to investigate the membrane and advanced
ozone membrane reactor performance for the KHP ozonation
reaction. Rate equations from semibatch reaction experiments
(Tables 1 and 2) were used along with the experimental ozone mass
transfer rate (Eq. (16)), membrane permeation rates (Eq. (17)) and
ozone solubility and diffusion data [58,59]. Fig. 4 shows that there
is good agreement between experimental data and model calcula-
tions. The model successfully captured the behavior of the reaction
in both membrane reactor configurations.
Membrane ozonation gave faster KHP conversion and better
Fig. 5. Calculated percent removals of KHP (solid lines) and TOC (dash lines) at
TOC removal compared to the semibatch reactor equipped with a different ozone dosage and ozone feed flow rates (QO3 ) ([KHP] = 250 ppmC, T = 25 ◦ C,
traditional glass sparger. This is mainly due to the higher ozone and  = 3.5 min).
126 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127
Fig. 6. Calculated percent removals of KHP (solid lines) and TOC (dash lines) as
a function of (a) membrane selectivity at different membrane fluxes ( = 3.5 min),
and (b) membrane areas ratio at different residence time ([KHP] = 250 ppmC,
[O3 ] = 130 ppm, QO3 = 20 sccm, and T = 25 ◦ C).
when ozone is fed at high concentration but low flow rate as high
ozone gas concentration and large interfacial area from small ozone
gas bubbles favor fast ozone mass transfer rate and reaction. Fig. 5
shows that high TOC removal is also obtained at this ozone feed
condition. The TOC removal has an asymptote of 60% at room tem-
perature due to the refractory nature of the I3 products. However
as most of the I3 products are carboxylic acids, it is possible to use
biological method for their removal.
The selective removal of water in the advanced ozone mem-
brane reactor concentrates the organics in the reaction zone
resulting in faster reaction and more complete degradation of
organics as shown by the twofold increase in TOC removal (Fig. 4b).
Model calculation was used to examine the effects of membrane
selectivity and flux on KHP and TOC removal in the advanced
ozone membrane reactor (Fig. 6a). The calculation shows that mem-
brane flux is more important than membrane selectivity, and that
a H2 O/organic separation of 10 is sufficient to enhance ozone
degradation of the organic pollutant. Increasing the membrane
selectivity from 10 to 100,000 resulted in less than 10% increase in
TOC removal, while a fourfold increase in membrane flux resulted
in over 20% improvement. The relative area of the membrane sep-
arator and distributor was examined in Fig. 6b. It is important to
note that the calculations were based on the tube-and-shell reac-
tor arrangement shown in Fig. 2 and therefore the membrane area
ratio is constrained to values less than 0.9. The plots show that for
all residence time, KHP and TOC removal increase as the area of the
separation membrane increases. This suggests that it would be ben-
eficial to consider capillary or spiral wound membranes to attain a
larger separation membrane area.
4. Concluding remarks
This work shows that an advanced ozone membrane reactor
that uses membranes for ozone distribution, reaction contactor and
water separation could attain both higher EDC and TOC removals
for KHP ozonation at a shorter residence time compared to a
traditional semibatch reactor. A mathematical model based on
experimental transport, permeation and reaction data gave an
accurate description of the advanced ozone membrane reactor
operation. The model was used to examine the reactor and reaction
to explore membrane properties, reactor design as well as oper-
ating conditions to achieve the best pollutant removal and TOC
degradation for the minimum amount of ozone use. Calculations
showed that the best result could be obtained from redesigning
the reactor to accommodate a larger separation membrane area.
The study also showed that for ozonation of KHP, it is possible to
achieve complete KHP conversion, but TOC removal approaches an
asymptote at 60% at room temperature due to the refractory nature
of the I3 products.
Acknowledgments
The authors gratefully acknowledge funding from the Hong
Kong Innovation Technology Fund (ITS/108/09FP) and financial
supports of Chiaphua Industries Ltd., Anjou Recherches and Veolia
Environment. We thank the Consulate General of France in Hong
Kong and the Foreign Ministry of France for financial support for
Ms. Angelique Blondy.
References
[1] G. Loraine, M. Pettigrove, Environmental Science and Technology 40 (2006)
687–695.
[2] B.L. Tan, D.W. Harker, J.F. Muller, F.D. Leusch, L.A. Tremblay, H.F. Chapman,
Chemosphere 69 (2007) 644–654.
[3] A.J. Watkinson, E.J. Murby, S.D. Costanzo, Water Research 41 (2007) 4164–4174.
[4] T. Okuda, Y. Kobayashi, R. Nagao, N. Yamashita, H. Tanaka, S. Tanaka, S. Fujii, C.
Konishi, I. Houwa, Water Science and Technology 57 (2008) 65–71.
[5] Y.J. Zhang, S.U. Geissen, C. Gal, Chemosphere 73 (2008) 1151–1161.
[6] D. Simazaki, J. Fujiwara, S. Manabe, M. Matsuda, M. Asami, S. Kunikane, Water
Science and Technology 58 (2008) 1129.
[7] D. Kolpin, E. Furlong, M. Meyer, E. Thurman, S. Zaugg, L. Barber, Environmental
Science and Technology 36 (2002) 1202–1211.
[8] M.J. Focazio, D.W. Kolpin, K.K. Barnes, E.T. Furlong, M.T. Meyer, S.D. Zaugg, L.B.
Barber, M.E. Thurman, Science of the Total Environment 402 (2008) 201–216.
[9] K. Kummerer (Ed.), Pharmaceuticals in the Environment – Sources, Fate, Effects
and Risks, 2nd ed., Springer, 2004.
[10] S.A. Synder, P. Westerhoff, Y. Yoon, D.L. Sedlak, Environmental Engineering
Science 20 (2003) 449–469.
[11] P.E. Stackelberg, E. Furlong, M. Meyer, S. Zaugg, A. Henderson, D. Reissman,
Science of the Total Environment 329 (2004) 99–113.
[12] P.E. Stackelberg, J. Gibs, E.T. Furlong, M.T. Meyer, S.D. Zaugg, R.L. Lippincott,
Science of the Total Environment 377 (2007) 255–272.
[13] Center for Disease Control and Prevention, Third National Report on Human
Exposure to Environmental Chemicals, National Center for Environmental
Health Pub. No. 05-0570, 2005.
[14] World Health Organization, Global assessment of the state-of-the-science
of Endocrine Disruptors, The International Programme on Chemical Safety,
WHO/PCS/EDC/03/3, 2003.
[15] T.E. Stoker, L.G. Parks, L.E. Gray, R.L. Cooper, Critical Reviews in Toxicology 30
(2000) 197–252.
[16] A.C. Collier, Ecohealth 4 (2007) 164–171.
[17] T.A. Ternes, M. Meisenheimer, D. McDowell, F. Sacher, H.J. Brauch, B.H. Gulde,
G. Preuss, U. Wilme, N.Z. Seibert, Environmental Science and Technology 36
(2002) 3855–3863.
[18] P. Westerhoff, Y. Yoon, S. Snyder, E. Wert, Environmental Science and Technol-
ogy 39 (2005) 6649–6663.
[19] V. Camel, A. Bermond, Water Research 32 (1998) 3208–3222.
[20] M.C. Huber, S. Canonica, G.Y. Park, U.V. Gunten, Environmental Science and
Technology 37 (2003) 5665–5670.
[21] R. Ermawati, S. Morimura, Y.Q. Tang, K. Liu, K. Kida, Journal of Bioscience and
Bioengineering 103 (2007) 27–31.
[22] Y.P. Zhang, J.L. Zhou, Water Research 39 (2005) 3991–4003.
[23] A.M. Comerton, R.C. Andrews, D.M. Bagley, C.Y. Hao, Journal of Membrane Sci-
ence 313 (2008) 323–335.
 H.L. Ho et al. / Catalysis Today 193 (2012) 120–127 127

[24] K. Kimura, S. Toshima, G. Amy, Y. Watanabe, Journal of Membrane Science 245 [42] J. Hoigne, H. Bader, Water Research 17 (1983) 173–183.
(2004) 71–78. [43] J. Hoigne, H. Bader, Water Research 17 (1983) 185–194.
[25] S.K. Hong, M. Elimelech, Journal of Membrane Science 132 (1997) 159–181. [44] L.T.Y. Au, J.L.H. Chau, C.T. Ariso, K.L. Yeung, Journal of Membrane Science 183
[26] S. van Geluwe, L. Braeken, B. van der Bruggen, Water Research 45 (2011) (2001) 269–291.
3551–3570. [45] L.T.Y. Au, W.Y. Mui, P.S. Lau, C.T. Ariso, K.L. Yeung, Microporous and Mesoporous
[27] B.S. Karnik, S.H.R. Davies, K.C. Chen, D.R. Jaglowski, M.J. Baumann, S.J. Masten, Materials 47 (2001) 203–216.
Water Research 39 (2004) 728–734. [46] W.C. Wong, L.T.Y. Au, P.P.S. Lau, C.T. Ariso, K.L. Yeung, Journal of Membrane
[28] B. Schlichter, V. Mavrov, H. Chmiel, Desalination 156 (2003) 257–265. Science 193 (2001) 141–161.
[29] P.V. Shanbhag, A.K. Guha, K.K. Sirkar, Industrial and Engineering Chemistry [47] Y.S. Cheng, M.A. Pena, J.L. Fierro, D.C.W. Hui, K.L. Yeung, Journal of Membrane
Research 37 (1998) 4388–4398. Science 204 (2002) 329–340.
[30] B.S. Karnik, S.H. Davies, M.J. Baumann, S.J. Masten, Environmental Science and [48] J.L.H. Chau, A.Y.L. Leung, K.L. Yeung, Lab-on-a-Chip 3 (2003) 53–55.
Technology 39 (2005) 7656–7661. [49] J.L.H. Chau, K.L. Yeung, Chemical Communications (2002) 960–961.
[31] K.L. Yeung, R. Aravind, J. Szegner, A. Varma, Studies in Surface Science and [50] J. Motuzas, S. Heng, P.P.S.Z. Lau, K.L. Yeung, Z.J. Beresnevicius, A. Julbe, Microp-
Catalysis 101 (1996) 1349–1358. orous and Mesoporous Materials 99 (2007) 197–205.
[32] M.A. Pena, D.M. Carr, K.L. Yeung, A. Varma, Chemical Engineering Science 53 [51] G.H. Yang, X.F. Zhang, S.Q. Liu, K.L. Yeung, J.Q. Wang, Journal of Physics and
(1998) 3821–3834. Chemistry of Solids 68 (2007) 26–31.
[33] K.L. Yeung, R. Aravind, R.J.X. Zawada, J. Szegner, G. Cao, A. Varma, Chemical [52] X.F. Zhang, H.O. Liu, K.L. Yeung, Materials Chemistry and Physics 96 (2006)
Engineering Science 49 (1994) 4823–4838. 42–50.
[34] Y.S. Cheng, M.A. Pena, K.L. Yeung, Journal of the Taiwan Institute of Chemical [53] S.M. Lai, L.T.Y. Au, K.L. Yeung, Microporous and Mesoporous Materials 54 (2002)
Engineers 40 (2009) 281–288. 63–77.
[35] Y. Guo, X.F. Zhang, H.Y. Zou, H.O. Liu, J.Q. Wang, K.L. Yeung, Chemical Commu- [54] S.M. Kwan, K.L. Yeung, Chemical Communications (2008)
nications (2009) 5898–5900. 3631–3633.
[36] A.J. Maira, W.N. Lau, C.Y. Lee, P.L. Yue, C.K. Chan, K.L. Yeung, Chemical Engineer- [55] S. Heng, P.P.S. Lau, K.L. Yeung, M. Djafer, J.C. Schrotter, Journal of Membrane
ing Science 58 (2003) 959–962. Science 243 (2004) 69–78.
[37] Y.S.S. Wan, K.L. Yeung, A. Gavriilidis, Applied Catalysis A 281 (2005) 285–293. [56] M.S. Eovitz, U. von Gunten, H.-P. Kaiser, Ozone Science and Engineering 22
[38] K.L. Yeung, X.F. Zhang, W.N. Lau, R. Martin-Aranda, Catalysis Today 110 (2005) (2000) 123–150.
26–37. [57] H. Benbelkacem, H. De bellefontaine, Chemical Engineering and Processing 42
[39] S. Heng, K.L. Yeung, M. Djafer, J.C. Schrotter, Journal of Membrane Science 289 (2003) 723–732.
(2007) 67–75. [58] J.A. Roth, D.E. Sullivan, Industrial and Engineering Chemistry Fundamentals 20
[40] S. Heng, K.L. Yeung, A. Julbe, A. Ayral, J.C. Schrotter, Microporous and Meso- (1981) 137–140.
porous Materials 115 (2008) 137–146. [59] P.N. Johnson, R.A. Davis, Journal of Chemical and Engineering Data 41 (1996)
[41] Y.H. Chen, N.C. Shang, D.C. Hsieh, Journal of Hazardous Materials 157 (2008) 1485–1487.
260–268.