Catalytic ozonation with γ-Al2O3 to enhance the degradation of refractory organics in water

Article abstract of DOI:10.1016/j.apcata.2014.10.037

Nowadays, heterogeneous catalytic ozonation appears as a promising way to treat industrial wastewaters containing refractory pollutants, which resist to biological treatments. Several oxides and minerals have been used and their behavior is subject to controversy with particularly the role of Lewis acid sites and/or basic sites and the effect of salts. In this study, millimetric mesoporous γ-Al₂O₃ particles suitable for industrial processes were used for enhancing the ozonation efficiency of petrochemical effluents without pH adjustment. A phenol (2,4-dimethylphenol (2,4-DMP)) was first chosen as petrochemical refractory molecule to evaluate the influence of alumina in ozonation. Single ozonation and ozonation in presence of γ-Al₂O₃ led to the disappearance of 2,4-DMP in 25 min and a decrease in pH from 4.5 to 2.5. No adsorption of 2,4-DMP occurred on γ-Al₂O₃. Adding γ-Al₂O₃ in the process resulted in an increase of the 2,4-DMP oxidation level. Indeed, the total organic carbon (TOC) removal was 14% for a single ozonation and 46% for ozonation with γ-Al₂O₃. Similarly, chemical oxygen demand (COD) removal increases from 35 to 75%, respectively. Various oxidized by-products were produced during the degradation of 2,4-DMP, but after 5 h ozonation 90% of organic by-products were acetic acid > formic acid ?

Full text of DOI:10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
Applied Catalysis A: General xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Catalytic ozonation with ␥-Al2O3 to enhance the degradation of
refractory organics in water
Jullian Vitteneta,b, Wael Aboussaoudc,d, Julie Mendreta, Jean-Ste´phane Picc,
Hubert Debellefontainec, Nicolas Lesagee, Karine Fauchere, Marie-He´le`ne Manerod,
Frédéric Thibault-Starzykf, Hervé Leclercf, Anne Galarneaub,∗, Ste´phan Brosillona
a IEM (Institut Européen des Membranes), UMR 5635 CNRS-ENSCM-UM2, Université Montpellier 2, Place E. Bataillon, 34095 Montpellier, France
b Institut Charles Gerhardt Montpellier, UMR 5253 CNRS/UM2/ENSCM/UM1–ENSCM, 8 rue de l’Ecole Normale, 34296 Montpellier Cedex 5, France
c Université de Toulouse, LISBP, UMR INSA/CNRS 5504 & UMR INSA/INRA 792, 135 avenue de Rangueil, 31077 Toulouse Cedex 4, France
d Université de Toulouse, INPT, UPS, Laboratoire de Génie Chimique, 4 allée Emile Monso, 31432 Toulouse Cedex 4, France
e TOTAL, PERL, RN117 BP 47, F-64170 Lacq, France
f
Laboratoire Catalyse et Spectrochimie, ENSICAEN Université de Caen CNRS, 6 Boulevard Maréchal Juin, 14050 Caen Cedex, France
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 15 July 2014
Received in revised form 13 October 2014
Accepted 15 October 2014
Available online xxx
Nowadays, heterogeneous catalytic ozonation appears as a promising way to treat industrial wastewa-
ters containing refractory pollutants, which resist to biological treatments. Several oxides and minerals
have been used and their behavior is subject to controversy with particularly the role of Lewis acid sites
and/or basic sites and the effect of salts. In this study, millimetric mesoporous ␥-Al2O3 particles suit-
able for industrial processes were used for enhancing the ozonation efficiency of petrochemical effluents
without pH adjustment. A phenol (2,4-dimethylphenol (2,4-DMP)) was first chosen as petrochemical
refractory molecule to evaluate the influence of alumina in ozonation. Single ozonation and ozonation in
presence of ␥-Al2O3 led to the disappearance of 2,4-DMP in 25 min and a decrease in pH from 4.5 to 2.5.
No adsorption of 2,4-DMP occurred on ␥-Al2O3. Adding ␥-Al2O3 in the process resulted in an increase of
the 2,4–DMP oxidation level. Indeed, the total organic carbon (TOC) removal was 14% for a single ozona-
tion and 46% for ozonation with ␥-Al2O3. Similarly, chemical oxygen demand (COD) removal increases
from 35 to 75%, respectively. Various oxidized by-products were produced during the degradation of
2,4-DMP, but after 5 h ozonation 90% of organic by-products were acetic acid > formic acid ꢀ oxalic acid.
Some of the carboxylic acids were adsorbed on ␥-Al2O3. The use of radical scavengers (tert-butanol)
highlighted the involvement of hydroxyl radicals during catalytic ozonation with ␥-Al2O3 in contrary
to single ozonation, which mainly involved direct ozone reaction. ␥-Al2O3 is an amphoteric solid with
Lewis acid AlOH(H+) sites and basicAl-OH sites. After ozonation the amount of basic sites decreased due
to carboxylates adsorption, while the Lewis acid sites remained constant as evidenced by FTIR. Several
ozonation runs with ␥-Al2O3 reported a progressive decrease of its catalytic activity due to the cumula-
tive sorption of carboxylates on the basic sites. After 80 h of ozonation, a calcination at 550 ◦C allowed
to recover allAl-OH basic sites and the initial activity of ␥-Al2O3. A synthetic petrochemical effluent
containing various petrochemicals (phenol, acetic acid, naphtenic acid, pyrene, naphtalene) was then
treated with ␥-Al2O3 with and without NaCl. Sodium ions prevented carboxylates adsorption on ␥-Al2O3
leading to a higher efficiency of ␥-Al2O3 in presence of NaCl and allowed to decrease the toxicity of the
petrochemical effluent.
Keywords:
Catalytic ozonation
Wastewater treatment
Al2O3
Phenol
Advanced oxidation process
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
In recent years, many efforts on the advanced oxidation
processes (AOPs) development have been done for wastewater
treatment especially in industries to treat organic compounds
resistant to biological treatments [1]. Among the AOPs, processes
∗
Corresponding author. Tel.: +33 4 67 16 34 68.
E-mail address: anne.galarneau@enscm.fr (A. Galarneau).
based on ozone appear as promising technologies for the removal
http://dx.doi.org/10.1016/j.apcata.2014.10.037
0926-860X/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
2
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
of refractory compounds [2–7]. In aqueous media, O3 can react
directly with organic compounds and/or indirectly when O3
is decomposed into free radicals, which have higher oxidizing
rate than O3 itself [2–4,8]. Consequently, favoring this second
route allows to reach higher levels of toxic pollutant mineral-
ization. The indirect reaction in single ozonation is influenced
by the pH of the solution (high pH), the solution composition
and the temperature, but it is limited to a certain extend of
mineralization [3,8]. In this context, some studies have focused
on heterogeneous catalytic ozonation in aqueous media since
the early 2000s, due to the enhancement of the production
of hydroxyl radicals with materials such as activated carbons,
zeolites, oxides (CeO2, TiO2, MgO, Al2O3, MnO2, Fe2O3, Co3O4,
NiO, CuO, ZnO, ZrO2,. . .) and the subsequent ozone economy
[4,8–17]. However, all of these materials have never been studied
in identical conditions, so it is difficult to compare their efficiency.
An excellent review by Nawrocki and Kassprzyk-Hordern in
2010 reported the different materials and conditions used in
catalytic ozonation in this last decade [8]. They report that the
following order in term of better heterogeneous catalysts appears
in certain conditions (ozonation of dinitrobenzene at pH 3):
carboxylic acids removal than single ozonation [22]. Recently,
Ikhlaq et al. [20] confirmed this hypothesis and pointed out
also that HO• radicals can combine with themselves to produce
H2O2, depending on the pH of the solution. This was demon-
strated after pH adjustment for pH higher than pH = 6.2. Another
catalytic reaction mechanism with aluminum oxides was also pro-
posed due to the ir capacity to adsorb ozone and to decompose
it into free radicals by interacting with the hydroxyl groups of
the materials [16,24]. Moreover, the catalytic effect of ␥-Al2O3
was preserved after several ozonation runs, indicating no poison-
ing of the active sites for ozone decomposition [16]. However,
the increase of the pollutant removal efficiency by heterogeneous
ozonation with ␥-Al2O3 was also attributed to the adsorption of
ozonation by-products. On the one hand, pharmaceutical pollutant
removal experiments were performed by alternating single ozona-
tion and catalytic ozonation with ␥-Al2O3 in order to observe the
by-products adsorption capacity of ␥-Al2O3. The authors explain
that carboxylates could be adsorbed onto ␥-Al2O3 when the solu-
tion pH is lower than the pHPZC of the material [2]. On the other
hand, 2,4-dimethylphenol (2,4-DMP) removal was performed by
single ozonation in order to generate oxidized by-products and,
Fe2O3 > Co3O4 > MoO3 > CuO ∼ NiO > Al2O3 > TiO2 > Cr2O3 ∼ MnO2 ∼ O3, then ␥-Al2O3 was added to examine its adsorption ability [17].
which seems to follow the basicity strength of the oxides
(Mn > Fe > Co > Ni > Cu > Zn) except for MnO2. However, in some
other studies MnO2 was demonstrated as the more active catalyst,
but leaching of some cations have been also observed, as well as
for CuO. Some of these oxides need to be supported for industrial
processes to avoid fine particles and the leaching of metal has to be
avoided. In this review was also highlighted the wide controversy
of mechanisms proposed in literature responsible for the increase
of ozone activity in the presence of materials and especially the
role of Lewis acid sites and/or basic sites, which remained a
subject of debate. Catalytic ozonation appears as a highly suitable
process for petrochemical wastewater treatments as ozone is
especially active for refractory molecules of this industry such
as aromatic molecules substituted with electron donor groups as
alkyl-benzene (toluene, xylene, ethylbenzene), phenols (alkylphe-
nols) and polyaromatic hydrocarbons (naphtalene, phenantrene,
anthracene, . . .).
It was previously shown that the catalytic activity of activated
carbons in ozonation decreased after several runs by the loss of
their basic sites and no regeneration was possible [5–7]. Inorganic
catalysts should be preferred. Zeolites and oxides in ozonation
for wastewater treatment provide many advantages: no chemi-
cals to add, simplicity of use and high pollutant removal efficiency
[15]. Concerning the zeolites in ozonation process for organic com-
pounds removal, it seems that the reaction mechanisms are zeolite
structures dependent [18]. Some studies revealed the adsorption
capacity of pollutant and/or ozone into the micropores of zeo-
lites, especially in ZSM-5 structure type, which does not confer
any radical pathway but serve as reservoir of ozone and adsorbents
of organic compounds [19–21]. Besides, it was also demonstrated
that zeolites could catalyze ozone decomposition and enhance the
generation of hydroxyl radicals (HO•) [11]. Indeed, it was recently
reported that basic zeolites as LTA zeolite can enhance radicals gen-
eration by combining two pathways: the production of hydroxide
ions during the cation-exchange of Na-zeolite with the protons of
the solution and by the interaction of ozone with some hydroxyl
groups present as defects in the material [18]. However, a decrease
of the LTA catalytic activity was observed for a second run of phe-
nol compounds ozonation due in part to the irreversibility of the
cation-exchange [18]. Other porous materials especially ␥-Al2O3
have been identified to generate free radicals from the interac-
tion between their hydroxyl groups and ozone [20,22,23]. A study
suggested that ␥-Al2O−3 combined t−o O3 allowed an enhancement
of HO• radical via •O2 and/or •O3 radicals, resulting in a faster
These experiments clearly evidenced no adsorption of the oxi-
dized by-products generated after 35 min of single ozonation,
but an adsorption of the by-products generated after 3 h of sin-
gle ozonation. Consequently, it appears that reaction mechanisms
in aluminum oxides/ozone processes are by-products dependent
and are still unclear, especially concerning ␥-Al2O3 material. The
adsorption of oxidized by-products and the role of Lewis acid sites
and basic sites of ␥-Al2O3 are subject to large controversy.
In the present study, a commercial ␥-Al2O3 with millimetric par-
ticles, suitable for industrial ozonation processes, was investigated
for 2,4-DMP degradation in the presence of ozone, without adding
any chemical, so without pH adjustment. Substituted phenols, such
as 2,4-DMP are highly toxic compounds and are typical pollutants
found in petrochemical wastewater [25,26]. The kinetics of pol-
lutant degradation was followed, as well as the formation of the
oxidized by-products by chromatography, by total organic carbon
(TOC) and chemical oxygen demand (COD) measurements. Adsorp-
tion isotherms of 2,4-DMP and of the final oxidized by-products
(carboxylic acids) on ␥-Al2O3 were performed to clarify the adsorp-
tion capacity of ␥-Al2O3. FTIR studies were performed to examine
the acid and basic sites of the material before and after ozonation.
The reusability of ␥-Al2O3 was investigated by performing several
ozonation runs of the pollutant. ␥-Al2O3 was also used in the ozona-
tion of a complex synthetic petrochemical effluent (phenol, acetic
acid, naphtenic acid, pyrene, naphtalene) to confirm the obser-
vations obtained with 2,4-DMP. Toxicity tests were performed to
evaluate the efficiency of ␥-Al2O3 in ozonation.
2. Experimental procedure
2.1. Materials
Acetonitrile (ACN) and water (H2O) for high performance liquid
chromatography (HPLC) analyzes are HPLC grade (Sigma Aldrich).
␥-Al2O3 was purchased from Alpha Degussa. All other chemicals
are high grade commercially available from Sigma Aldrich: 2,4-
dimethylphenol (2,4-DMP) (98%), acetic acid, formic acid, oxalic
acid, phenol, pyrene, naphtalene and tert-butanol (t-BuOH). Naph-
tenic acid was purchased from Fluka. 2,4-DMP and carboxylic acids
solutions were prepared using Millipore Milli-Q water.
2.2. Inorganic materials characterization
X-ray diffraction (XRD) pattern of the material was performed
using a Bruker D8 Advance diffractometer with a Bragg-Brentano
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
3
geometry and equipped with a Bruker Lynx Eye detector. XRD pat-
tern was recorded in the range 4◦–85◦ (2ꢀ) with a 0.0197◦ angular
step size and a 0.2 s counting time per step. Textural properties
of the material were determined by N2 adsorption/desorption at
–196 ◦C on a Micromeritics Tristar 3000 apparatus. The sample was
previously outgassed in vacuum at 250 ◦C for 12 h. The surface area
was determined according to the BET method. Total pore volume
was calculated at the end of the adsorption step. Mesopore diame-
ter was evaluated using the Broekhoff and De Boer method applied
to the desorption branch of the isotherm as it has been shown to
be one of the more accurate method for mesopore size evaluation
[27].
These experiments were performed using another lab scale reactor
described elsewhere [17], similar to the first one but with a volume
of solution of 2 L, an amount of ␥-Al2O3 of 5 g L−1, a O3 gas flow rate
is 30 L h−1 and O3 gas concentration is 5 g Nm−3 (obtained from O2).
A synthetic petrochemical effluent was also ozonized with ␥-
Al2O3 without and with NaCl (50 g L−1). These experiments were
performed in a third pilot plant scale reactor described else-
where [28] with a volume of solution of 1.5 L, an amount of
␥-Al2O3 of 2 g L−1, a O3 gas flow rate of 24 L h−1 and a gas con-
centration of 5 g O3 Nm−3. The complex synthetic effluent was a
mixture of several aromatic hydrocarbons and associated acids:
phenol, acetic acid, naphtenic acid, pyrene, naphtalene with ini-
tial concentration of 200, 200, 25, 0.05, 0.95 mg L−1, respectively,
corresponding to TOC = 230 mg L−1 and COD = 750 mg L−1. Toxicity
tests were performed on the experiments performed in presence
of NaCl (50 g L−1). Non-standardized toxicity tests (Toxkits), named
ARTOXKIT and ROTOXKIT, were performed using two different sea-
water organisms Artemia franciscana and Brachionus calyciflorus,
respectively. Toxicity tests were performed after pH adjustment at
pH 7 of the effluent treated by 150 and 300 min of ozonation with
␥-Al2O3. The efficiency of the process was assessed by the effluent
concentration able to provoke 50% mortality of the species after
24 h of exposure (LC50 24 h). Results were expressed in toxic unit:
Equitox/m3 = 100/LC50 24 h.
The 13C MAS NMR spectrum of ␥-Al2O3 after 5 h ozonation was
performed on filtered and dried material and was obtained on a Var-
ian VNMRS 300 MHz spectrometer with a magnetic field of 7.05 T.
The operating frequency for 13C was 75.43 MHz. The sample was
packed in a 7.5 mm ZrO2 rotor. The chemical shifts are presented
in parts per million.
FTIR analyses of ␥-Al2O3 were performed on dried materials
before and after ozonation. Samples were pressed (∼106 Torr) into
self-supported disks (2 cm2 area, 7–10 mg cm−2) and placed in a
quartz cell equipped with KBr windows. A movable quartz sam-
ple holder permits one to adjust the pellet in the infrared beam
for spectra recording and to displace it into a furnace at the top of
the cell for thermal treatment. The cell was connected to a vacuum
line for evacuation (Presidual ≈ 10−6 Torr) and for the introduction
of gases into the infrared cell. The gas pressure inside the cell was
measured by a pressure gauge (10−2–103 Torr range). A Bruker
Vertex 80 V spectrometer equipped with a mercury cadmium tel-
luride (MCT) cryodetector and an extended KBr beam splitter was
used for the acquisition of spectra recorded at room tempera-
ture, in the 600–5500 cm−1 range. The resolution of the spectra
was 4 cm−1, and 128 scans were accumulated for each spectrum.
Strength of acid sites of the materials was obtained by pyridine
adsorption followed by FTIR spectroscopy for different tempera-
tures of desorption. Basic sites of the materials were analyzed by
CO2 adsorption/desorption.
2.4. Analysis of soluble pollutant and by-products
The 2,4-DMP concentration in the aqueous solutions was
monitored by HPLC (Waters 600E controller and pump, 717
plus Autosampler) equipped with an UV detector (2996 PDA,
fixed wavelength ꢁ = 279 nm). HPLC analyzes was performed on
a C18 grafted silica column (Interchim, Uptisphere® C18-ODB,
150 mm × 4.6 mm × 5 ␮m). 20 ␮l sample were injected and
a
mobile phase composed of 45/55 (ACN/H2O) at 0.8 ml min−1 was
used. Solutions were filtered before analysis with a PTFE syringe fil-
ter (0.45 ␮m). Under these analytical conditions, the retention time
of 2,4-DMP was about 5 min.
The total organic carbon (TOC) values of the solutions were
determined with a Shimadzu TOC-V meter. The gas flow-rate (air)
was set to 130 mL min−1 and the gas pressure was 190 kPa. The
TOC measurement is based on the oxidation of the organic mat-
ter (except volatile organic compounds) by thermal oxidation. The
solution is first acidified with HCl (2 N) to transform carbonates into
CO2. Air is then bubbled in the solution to remove CO2. Samples are
then injected in an oven heated at 680 ◦C containing a catalyst (Pt)
to transform the organics into CO2, which is therefore analyzed by
an IR detector.
The chemical oxygen demand (COD) is the quantity of oxygen
necessary to chemically oxidize all molecules in water. This analysis
is performed using oxidation with K2Cr2O7 in closed containers in
acidic medium in presence of Ag2SO4 catalyst. HgSO4 is also added
to precipitate chloride anions. After addition of 2 mL of polluted
water solution in tubes containing the oxidative reactants at dif-
ferent concentrations to analyze COD between 0 and 150 mgO2 L−1
(purchased to Hash), the mixtures were heated at 150 ◦C for 2 h.
The amount of produced Cr3+ was measured by spectrophotometry
(Hash instrument) at 600 nm, which allowed to calculate COD.
The carboxylic acids concentrations were determined by ionic
chromatography (IC) using a Dionex ICS 1000 system equipped
with an ASRS 4 mm ionization system and a Dionex AS19 col-
umn. 25 ␮l samples were injected. Eluent was a KOH solution at
1 mL min−1 and with the following gradient conditions: 10 mM
from 0 to 10 min, then 45 mM from 10 to 40 min and finally 10 mM
from 40 to 50 min.
2.3. Adsorption and ozonation experiments
Prior to ozonation experiments, 2,4-DMP adsorption capacity
of ␥-Al2O3 was assessed to quantify its possible contribution to
the degradation of the pollutant. Adsorption of the pollutant was
performed at 25 ◦C in abiotic conditions: 0.2 g of ␥-Al2O3 were
stirred in a glass bottle containing 100 mL of pollutant solution
until the equilibrium was reached, which was corresponding to 4 h.
Equilibrium isotherms of adsorption were plotted for 2,4-DMP con-
centrations ranging from 1 to 400 mg L−1. Equilibrium isotherms of
carboxylic acids adsorption (acetic acid, formic acid, oxalic acid)
were performed following the same procedure than for 2,4-DMP
adsorption.
The ozone decomposition of the pollutant was assessed in a glass
stirred batch reactor of 2 L as previously described [18]. Experi-
ments were conducted at 25 ◦C, feeding the reactor with a 40 L h−1
O3 gas flow rate with a 2 g Nm−3 O3 concentration (obtained from
air). In a typical run, 3 g of ␥-Al2O3 were added in the reactor with
1.5 L of 2,4-DMP solution at 50 mg L−1. The solution was stirred
at 400 rpm at 25 ◦C and O3 was injected to the reactor immedi-
ately after introducing ␥-Al2O3. In order to avoid limitation by O3
gas to liquid transfer, the stirring velocity of the reactor was opti-
mized by progressively increasing the velocity from 200 to 400 rpm
and optimal stirrer velocity was found at 400 rpm [18]. The pollut-
ant removal and its oxidized by-products were monitored taking
samples within time through PTFE syringe filters (0.45 ␮m).
The reuse and regeneration of ␥–Al2O3 were also investigated
by performing several ozonation runs of the 2,4-DMP degradation.
LC-MS analyses were achieved with a UPLC Acquity H-Class
(Waters) and a Synapt G2-S (Waters) detector equipped with a
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
4
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
Fig. 1. N2 adsorption (•) and desorption (ꢁ) isotherms of ␥-Al2O3 at −196 ◦C.
dual electro-spray (EScl). 10 ␮L samples were injected, a column
Acquity BEH I was used (C18, 50 mm × 2.1 mm × 1.7 ␮m) and the
mobile phase was constituted of ACN and H2O at 0.8 ml min−1
3. Results and discussion
.
3.1. Characterization of ꢂ-Al2O3
␥-Al2O3 was purchased as 3–5 mm cylindrical pellets (Fig. S1),
which is a suitable size for industrial ozonation processes. XRD pat-
tern indicated principally three broad peaks around 35, 45 and 66◦
in 2ꢀ (Fig. S2), which showed the gamma crystalline phase of this
alumina oxide [29]. N2 adsorption/desorption isotherm of ␥-Al2O3
is of type IV according to IUPAC classification characteristic of a
mesoporous material (Fig. 1). A hysteresis loop (H1) was observed
at high relative pressures revealing cylindrical pores of mean pore
diameter of 15 nm. The material featured a BET specific surface area
Fig. 2. 2,4-DMP removal (a) and TOC removal (b) during single ozonation (ꢀ) and
during ozone/␥-Al2O3 coupled treatment with different amount of ␥-Al2O3: 1 g L−1
(+); 2 g L−1 (•); 5 g L−1 (*). Temperature: 25 ◦C. O3 flow rate: 40 L h−1. [O3]Inlet: 2 g
Nm–3. [2,4–DMP]0: 50 mg L–1. Volume of solution: 1.5 L.
of 236 m2 g−1 and a mesopore volume of 0.60 cm3 g−1
.
(Fig. 2) and the chemical oxygen demand (COD) (Fig. 3; Table 1)
with and without ␥-Al2O3. Some 2,4-DMP oxidized by-products
were identified and followed within time (Table 2).
3.2. Adsorption of 2,4-DMP on ꢂ-Al2O3
Prior to O3 experiments, adsorption of the pollutant on ␥-Al2O3
was verified to assess the adsorption contribution in the 2,4-DMP
removal. First, adsorption kinetic was monitored at 25 ◦C by stir-
ring a 2,4–DMP solution (50 mg L−1) with ␥-Al2O3 (2 g L–1) in order
to estimate the time necessary to reach equilibrium. The expected
maximum adsorption would be 25 mg 2,4-DMP per gram ␥-Al2O3.
Equilibrium was reached after 4 h and only 0.2 mg 2,4-DMP per
gram ␥-Al2O3 was obtained. The 2,4-DMP adsorption isotherm was
then built for an equilibrium time of 4 h and by varying the 2,4-DMP
concentration from 1 to 400 mg L−1. For the highest 2,4-DMP con-
centration (400 mg L−1) only 3.2 mg 2,4-DMP were adsorbed per
gram of ␥-Al2O3 (Fig. S3). Therefore, in this study (using a 2,4-DMP
solution of 50 mg L−1) the contribution of the pollutant adsorption
in the ozonation process can be neglected.
3.3.1. Removal of 2,4-DMP from the water solution
By single ozonation (without materials) 2,4-DMP was totally
removed from the solution after 25 min even with low ozone con-
centration (2 g Nm−3) (Fig. 2a). Similar fast 2,4-DMP removal by
single ozonation was also previously noticed with similar initial
concentration indicating a high reactivity of this pollutant with
ozone [17,18,30]. Adding ␥-Al2O3 (2 g L−1) in the ozonation process
led to a slight increase of 2,4-DMP removal rate with a complete
removal after 22 min (Fig. 2a), as observed also previously using
zeolites in similar conditions [17,18]. The initial degradation of 2,4-
DMP seems independent of the presence of any inorganic catalysts
and encountered presumably to fast direct reaction of ozone with
2,4-DMP. Ozone reacts preferentially with organic molecules fea-
turing double bounds (Scheme 1) and activated aromatics. 2,4-DMP
is an aromatic molecule substituted by three electron donor groups
(one hydroxyl strongly e− donor and two methyl) and is therefore
very favorable to ozonation degradation. Two simplified path-
ways can be imagined for direct ozonation of 2,4-DMP due to the
3.3. Ozonation of 2,4-DMP in aqueous solution with and without
ꢂ-Al2O3
Ozonation experiments were conducted with 2,4-DMP solu-
tions (50 mg L–1) at 25 ◦C with an initial pH of deionized water
around 4.5. No chemicals were added in the solution in accordance
of industrial process requirement (no buffer or base added). The
pH was continuously measured but not controlled. The pH value of
the 2,4-DMP solutions with and without ␥-Al2O3 was monitored
during the ozonation experiments (Fig. S4). The pollutant removal
over time was monitored as well as the total organic carbon (TOC)
Table 1
Rate constants for COD removal during 2,4-DMP ozonation for single ozonation and
ozonation with ␥-Al2O3.
kCOD fast × 105 (s−1
)
kCOD slow × 105 (s−1
)
Single ozonation
O3/␥-Al2O3
7.8 0.1
11.9 0.1
1.1 0.1
5.8 0.1
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
5
Fig. 3. COD removal (Inset: logarithm representation for rate constant calculation) (a) and AOSC factor evolution (b) during single ozonation (ꢀ) and during ozone/␥–Al2O3
coupled treatment (᭹). Temperature: 25 ◦C; O3 flow rate: 40 L h−1; [O3]Inlet: 2 g Nm−3; [2,4-DMP]0: 50 mg L−1; volume of solution: 1.5 L; ␥-Al2O3: 2 g L−1
.
polarizability of the double bond bearing the OH group (Scheme 2).
Different intermediary by-products are however expected as for
nitrobenzene ozonation [31]. Expected final products are car-
boxylic acids. They are very stable against ozonation as acetic acid
but not formic acid, which reacts with ozone to give CO2.
3.3.2. Removal of total organic carbon from the water solution
The removal of specific pollutant (as 2,4-DMP in our case) is
one of the objectives of the water treatment. However, the effi-
ciency of a water treatment process should also assess the amount
of total organic carbons (TOC) remaining in the solution result-
ing from the degradation of the initial product. Indeed oxidized
by-products of pollutants may have also a toxic character and
sometimes are even more toxic than the initial pollutant, as it will
be shown in the last part of this study. So, it is of prime impor-
tance to estimate the oxidation efficiency of a water treatment
process through TOC measurement [32,33]. If no adsorption of pol-
lutants or by-products is noticed, TOC removal corresponds to the
mineralization level of the pollutant. Concerning single ozonation,
only 14% TOC removal was reached after 5 h ozonation. Adding
Scheme 1. Schematic representation of direct O3 reaction with unsaturated bond.
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
6
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
efficiencies for 2 and 5 g L−1 ␥-Al2O3 were overlapping during
90 min of reaction. We may explain this fact considering two par-
allel reactions: catalytic ozonation and adsorption on ␥-Al2O3 of
highly oxidized by-products formed after 90 min ozonation. Similar
catalytic ozonation efficiency between 2 and 5 g L−1 ␥-Al2O3 is sus-
pected, whereas adsorption ability of highly oxidized by-products
is enhanced with higher amount of ␥-Al2O3 adsorbent.
Table 2
Identification by LC-MS of some 2,4-DMP oxidized by-products after single ozona-
tion and ozonation with ␥-Al2O3.
2,4-DMP oxidized
by-products
m/z
O3
O3/␥-Al2O3
tr (min)
tr (min)
td (min)
td (min)
3.3.3. Chemical oxygen demand (COD) removal
The chemical oxygen demand (COD) is another way to indirectly
measure the amount of organic compounds remaining in water.
TOC measures the organic matter that can be oxidized by thermal
treatment (expressed in mg of C per L), whereas COD measures the
organic matter that can be oxidized chemically. COD is expressed in
mg of O2 per L, which indicates the mass of oxygen consumed per
liter of solution to oxidize organics [11]. Single 2,4-DMP ozona-
tion led to 35% COD removal corresponding to a decrease from
130 to 83 mgO2 L−1 after 5 h treatment. Similarly to the TOC evo-
lution, the COD decreased more rapidly in the presence of ␥-Al2O3.
Indeed, adding ␥-Al2O3 (2 g L−1) in the process allowed a signifi-
cant reduction of the COD from 143 to 37 mgO2 L−1, corresponding
to 75% COD removal. The kinetics of COD removal was modeled
using a two-pathways first-order kinetic model [2]. This model dis-
tinguishes organic compounds oxidizing fast during the ozonation
experiments from those oxidizing slowly. The experimental data
for fast and slow oxidations were fitted with the first order kinetic
model (Eq. (1)) and reported in Fig. 3a. In each pathway of the model
the corresponding rate constants for COD removal were higher in
the presence of ␥-Al2O3 (Table 1). The change of regime was clearly
observed after 60 min of ozonation (Fig. 3a).
139
171
1.30
0.70
61
91
1.35
0.71
61
91
OH
O
C
C
HO
C8H10O4
C8H8O
121
1.27
91
1.32
85
O
OH
HO
C10H10O5
211
0.71
91
0.71
91
d[COD]x
= k[COD]x
dt
tr: retention time; td: complete removal time.
d[COD]x
[COD]x
(1)
␥-Al2O3 (2 g L−1) to the ozonation process allowed a remarkable
increase of the TOC removal corresponding to 46% of organic car-
bons removal (Fig. 2b). TOC removal increased linearly during
about 60 min and then became slower (especially after ca. 170 min),
which probably indicated that some oxidized by-products are very
stable towards ozonation. Moreover, this slowdown in the miner-
alization rate can be explained by the low ozone dosage chosen in
this study (2 g Nm−3) in comparison with usual industrial condi-
tions (25 to 100 g Nm−3). Increasing the amount of ␥-Al2O3 from
1 to 5 g L−1 in the ozonation process yielded to an increase of
TOC removal from 26 to 57%, respectively. However, TOC removal
ꢀ
CODx
COD0
− ln
with x = fast or slow.
An interesting factor to consider is the average oxidation state of
carbon (AOSC) [2,34,35], which informs indirectly about the aver-
age oxidation state of the organic compounds remaining in the
solution by comparing thermal oxidation (TOC) with chemical oxi-
dation (COD) according to Eq. (2).
ꢀ
ꢁ
COD
TOC
AOSC = 4 − 4 × 0.375
(2)
with 0.375 being the ratio between the molecular mass of C and O2
to expressed COD and TOC in mmol.
Higher AOSC values will be obtained for higher oxygenated
molecules. AOSC values are in the range [−4 to +4] with −4 account-
ing for low oxidation state as methane and +4 for high oxidation
state as CO2. The AOSC value of the starting 2,4-DMP solution is
around −1.4, as the average oxidation state of the molecule is low.
For single ozonation, the average oxidation level of the organic
compounds increased to −0.2 after 4 h indicating an increase of the
average oxidation states of the organics in solution, which how-
ever remains low. Adding ␥-Al2O3 during the ozonation process
resulted in a three steps process: from 0 to 60 min (AOSC from −1.4
to −0.2), from 60 to 170 min (AOSC = −0.2) and above 170 min with
a remarkable increase of the AOSC from −0.2 to +1.6 between 3
and 5 h ozonation (Fig. 3b). This evidences that the by-products
obtained after 170 min of ozonation with ␥-Al2O3 are different to
the one obtained by single ozonation and have a higher level of
oxidation state.
Scheme 2. Schematic representation of expected direct O3 reaction with 2,4-DMP
(plain arrows). Possible dimerization route (6) and methylmalonic acid (7) formation
is presented in dashed line.
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
7
toxicity level of the solution [32,33]. The identification of the first
by-products from 2,4-DMP oxidation during single ozonation was
attempted by LC-MS analysis. The non-exhaustive list of the prod-
ucts observed is presented in Table 2. Considering the products
identified, several types of reactions were involved in the oxida-
tion process:(i) Hydroxylation of the aromatic ring: hydroxylation
of the aromatic ring, producing C8H10O2 (product 1 in Scheme 2)
is consistent with hydroxylation products usually observed during
phenol ozonation [40,41]. This by-product was the most abundant
(Fig. S5). It appeared since the beginning of the reaction, reached
an optimum after 25 min ozonation and was totally degraded after
61 min of reaction (Table 2).(ii) Opening of the aromatic ring by
direct addition of ozone (Criegee mechanism): the identification of
C8H10O4 (product 2 in Scheme 2) confirms that the direct addition
of ozone on 2,4-DMP led to opening the aromatic ring and pro-
duced dimethyl muconic acid. Indeed, muconic acid is considered
as one of the major transient products during ozonation of phenol
[41,42]. This transient product was at its maximum in 20 min and
was totally degraded after 91 min (Fig. S5; Table 2).(iii) Dimeriza-
tion of 2,4-DMP: during the first minutes of ozonation the solution
became turbid indicating the apparition of less soluble products.
The compound C10H10O5 (Table 2; Fig. S5) has a higher molar mass
than 2,4-DMP, which could result from the reaction of oxalic acid
with C8H10O2 (product 1 in Scheme 2) and/or a degradation of a
dimer. Dimerization of phenoxy radicals in acidic water was pre-
viously observed leading to non-soluble species [30,43]. Such a
dimerization of dimethylphenols was also observed in oxidizing
conditions in presence of ozone or copper [41,44,45]. Another com-
pound C8H8O observed by LC-MS is considered as a by-product of
Fig. 4. TOC removal during single ozonation with (♦) and without (ꢀ) t-BuOH radical
scavenger and during ozone/␥-Al2O3 coupled treatment with (ꢁ) and without (•)
t-BuOH. Temperature: 25 ◦C; O3 flow rate: 40 L h−1; [O3]Inlet: 2 g Nm−3; [2,4-DMP]0:
50 mg L−1; volume of solution: 1.5 L; ␥-Al2O3: 2 g L−1; [t-BuOH]: 25 mg L−1
.
3.4. Ozonation of 2,4-DMP in presence of HO• radical scavenger
(t-BuOH)
The pH value of the 2,4-DMP solutions with and without ␥-
Al2O3 was monitored during the ozonation experiments (Fig. S4).
For single ozonation, a decrease in the pH value was observed from
4.0 to 2.8. This is very likely due to the formation of carboxylic
acids during ozonation as reported elsewhere [36–39]. A higher pH
decrease was noticed for ozonation in presence of ␥-Al2O3 from
4.5 to 2.5 after 5 h. In both cases (with and without ␥-Al2O3) the
promotion of HO• radicals due to basic pH is not possible due
to very low OH− concentration at the pH of this study. As men-
tioned previously, O3 can react by two routes in aqueous media:
the direct O3 ozonation (Schemes 1 and 2) and the radical type
reaction. This later route implies the generation of HO• radicals dur-
ing ozone self-decomposition. The radical reaction is more efficient
for removing refractory compounds [3,8]. In order to evidence HO•
generation in the ozonation with ␥-Al2O3 a stable radical scavenger
as tert-butanol (t-BuOH at 25 mg L−1) was introduced in the 2,4-
DMP solution. This amount of t-BuOH was previously demonstrated
to be enough in heterogeneous catalytic ozonation to annihilate
radical type reaction [40]. Single ozonation in presence of t-BuOH
revealed a slight decrease in the TOC removal efficiency from 14
to 11% (Fig. 4) showing that single ozonation was principally gov-
erned by direct reaction of ozone with the organics present in the
solution. For ozonation in the presence of ␥-Al2O3, adding t-BuOH
induced a decrease of the TOC removal from 46 to 30%. This high-
lighted that ␥-Al2O3 promotes the formation of HO• as no t-BuOH
was adsorbed on the material (not shown). As reported in sev-
eral studies [16,20,22,24], the aluminum oxides might enhance free
radicals generation due to their surface hydroxyl groups, which can
interact with O3. However, as TOC removal in presence of t-BuOH
remained higher (30%) with ␥-Al2O3 in comparison to single ozona-
tion (11%), it can be reasonably assumed that adsorption of oxidized
by-products occurred on ␥-Al2O3 and contributed to decrease the
TOC of the solution. If the direct ozone reaction is at the same level
with ␥-Al2O3 than for single ozonation therefore 11%, the contri-
bution of adsorption in TOC removal for ␥-Al2O3 could be 19% and
the contribution of radical reaction 16%.
C10H10O5 after several successive demethylations of the ring and
decarboxylation of the aliphatic chain bonded with the first carbon
of the aromatic ring. These two by-products were at their maximum
at 15 min and were completed degraded after 91 min of ozonation
(Fig. S5; Table 2).
3.5.2. By-products identified by LC-MS during ozonation in the
presence of ꢂ-Al2O3
During the degradation of the pollutant in presence of ␥-Al2O3,
LC-MS analyzes revealed the same by-products than for the single
ozonation and thus the same oxidation pathways. The maximum
apparition and the complete removal of these by-products occurred
at the same time than for single ozonation. All of these by-products
have been degraded after 90 min (Table 2; Fig. S5). The main differ-
ence consists in the amount of these by-products, which is much
abundant in presence of ␥-Al2O3 (except for C10H10O5, which was
formed in similar amount) (Fig. S5).
3.5.3. Identification and quantification of carboxylic acids
obtained during single ozonation
Ionic chromotography analysis allowed to identify and quan-
tify the carboxylic acids formed during 2,4-DMP ozonation. Three
carboxylic acids were mainly formed in the following increasing
amount (Fig. 5a): acetic acid > formic acid ꢀ oxalic acid (products
3, 5, 4 in Scheme 2). Acetic acid was the most abundant reaching
a plateau after 100 min, which corresponded to a final concentra-
tion of 62 mg L−1. The formation of acetic acid is a consequence
of an opening of the aromatic ring that could occur from the
beginning of the ozonation reaction and of the oxidation of the
by-products (Scheme 2). It must be noticed that acetic acid is
a highly refractory molecule that is very stable against ozona-
tion and will not be oxidized in formic acid or oxalic acid. The
release of formic acid can correspond to a demethylation of 2,4-
DMP from the beginning of the reaction and/or to the oxidation of
other by-products after several breakages of the C–C bonds. The
highest formic acid concentration (28 mg L−1) was observed after
60 min of reaction and then decreased to reach a concentration
3.5. Identification of by-products obtained during 2,4-DMP
ozonation
3.5.1. By-products identified by LC-MS during single ozonation
As oxidized phenolic compounds are known to have a possible
toxic character, their identification can give information about the
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
8
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
(Scheme 2). The higher amount of acetic acid and the lower amount
of formic acid were presumably due to the route implying dimer-
ization of 2,4-DMP. The very low amount of oxalic acid could result
from its reaction with the bihydroxylated by-product (product 1 in
Scheme 2) as soon as it was produced. Taking into account TOC mea-
surement and the initial amount of 2,4-DMP (3.28 mmol C L−1) it
was possible to determine the amount of organics remaining in the
solution and the level of mineralization (Table 3). After 5 h of single
ozonation 2.82 mmol C L−1 remained in the solution composed at
90% of carboxylic acid and 10% of unknown by-products.
3.5.4. Identification and quantification of carboxylic acids
obtained during ozonation with ꢂ-Al2O3
During ozonation with ␥-Al2O3 the same carboxylic acids as for
single ozonation were identified and quantified: 1 molecule of 2,4-
DMP was converted into 1.7 molecules CH3COOH, 1.0 molecule
HCOOH and 0.05 molecule HOOC-COOH. Lower amounts of car-
boxylic acids were obtained for ozonation with ␥-Al2O3 except for
oxalic acid, which followed the same trend as for single ozonation
(Fig. 5b). By allowing the generation of HO• radicals ␥-Al2O3 favors
a different oxidation mechanism leading to less acetic acid and less
formic acid. It could also be hypothesized that a non-reactive phe-
nomenon occurred such as the adsorption of the carboxylic acids
on the solid. Indeed the maximum concentration of acetic acid
was reached after 75 min and then decreased, which cannot be
due to the oxidation of acetic acid. Adsorption of carboxylic acids
was thus investigated in the following paragraph. Concerning mass
balance calculated from TOC and from the amount of carboxylic
acids (Table 3) it was demonstrated that ozonation in presence of
␥-Al2O3 generated after 1 h ozonation more by-products (other
than carboxylic acids) due to the different mechanism path-
way using HO• radicals with 0.73 mmol Cby-product L−1 after 1 h
for ozonation with ␥-Al2O3, whereas single ozonation produced
Fig. 5. Kinetic of acetic acid (ꢂ), formic acid (×) and oxalic acid (ꢁ) formation during
single ozonation (a) and during ozone/␥-Al2O3 coupled treatment (b). Temperature:
25 ◦C; O3 flow rate: 40 L h−1; [O3]Inlet: 2 g Nm−3; [2,4-DMP]0: 50 mg L−1; volume of
only 0.43 mmol Cby-product L
−1. After 5 h ozonation the amount of
solution: 1.5 L; ␥-Al2O3: 2 g L−1
.
by-products (other than carboxylic acids) decreased and reached
a similar value of ∼0.26 mmol Cby-product L−1 for both ozonation
processes. However, the remaining by-products (other than car-
boxylic acids) resulting from 5 h ozonation with ␥-Al2O3 are
different from those obtained by single ozonation as they featured
a higher level of oxidation state as previously demonstrated from
AOSC measurements. Carboxylic acids represent 86% of the carbon
remaining in the solution after 5 h ozonation. Ozonation with ␥-
Al2O3 allowed after 5 h to decrease the amount of organics in water
from 3.28 to 1.77 mmol C L−1, whereas single ozonation reached
of 21 mg L−1 after 5 h. This decrease is due to its oxidation into
carbon dioxide (Scheme 2). Concerning oxalic acid, a slight and
steady increase of its concentration was noticed to reach a value
of 6 mg L−1 after 5 h ozonation. Oxalic acid can be released from an
opening of the aromatic ring and/or from the oxidation of dimers.
The study of mass balance of the process (Table 3) indicated that
after 1 h of reaction 0.41 mmol of 2,4-DMP was transformed into
1 mmol CH3COOH, 0.65 mmol HCOOH and 0.02 mmolHOOC-COOH,
which corresponds to the transformation of 1 molecule of 2,4-DMP
into 2.4 molecules of CH3COOH, 1.6 molecules of HCOOH and 0.05
molecule of HOOC-COOH. This result is close to the expected sim-
plified direct ozonation of 2,4-DMP into acetic acid and formic acid
only 2.82 mmol C L−1
.
3.6. Adsorption of carboxylic acids on ꢂ-Al2O3
3.6.1. Adsorption isotherms of carboxylic acids on ꢂ-Al2O3
During 2,4-DMP oxidation, three carboxylic acids (acetic acid,
formic acid and oxalic acid) have been identified and ␥-Al2O3
adsorption properties towards these acids have been studied. The
adsorption isotherms of the three carboxylic acids were carried
out varying the acids concentrations from 5 to 400 mg L−1 (Fig. 6).
The adsorption capacity of ␥-Al2O3 became significant for equi-
librium acid concentration higher than 25 mg L−1. For the highest
acid concentration (400 mg L−1) an adsorption 20, 23, 35 mg g−1
(or 0.33, 0.50 and 0.38 mmol g−1) was obtained for acetic acid,
formic acid and oxalic acid, respectively. During ozonation with ␥-
Al2O3 the maximum acid equilibrium concentrations were found
after 1 h and corresponded to 43, 20 and 2 mg L−1, for acetic acid,
formic acid and oxalic acid, respectively. Considering the adsorp-
tion isotherms, an adsorption capacity on ␥-Al2O3 of 3.4, 3.7
and 0.9 mg g−1 (or 0.06, 0.08 and 0.01 mmol g−1) for acetic acid,
formic acid and oxalic acid, respectively, could be expected, which
will contribute as 0.43 mmol C L−1 in TOC removal. Adsorption of
Table 3
Mass balance after 1 and 5 h of single ozonation and ozonation with ␥-Al2O3 of
2,4-DMP.
O3
O3 + ␥-Al2O3
Ozonation time
1 h
5 h
1 h
5 h
CH3COOH (mmol L−1
)
1
1
0.71
0.43
0.02
0.73
0.60
0.24
0.04
0.25
HCOOH (mmol L−1
)
0.65
0.02
0.43
0.45
0.05
0.27
HOOC-COOH (mmol L−1
Other by-products
)
(mmol C L−1
)
Total by-products (mmol C L−1
Produced CO2 (mmol C L−1
Produced CO2 + adsorbated
by-products (mmol C L−1
)
3.12
0.16
2.82
0.46
2.62
0.66
1.77
1.51
)
)
Initial concentration 2,4-DMP: 50 mg L−1 (0.41 mmol L−1; 3.28 mmol C L−1).
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
9
giving peaks just higher than the background noise on the spec-
trum.
3.6.3. Identification of carboxylates adsorbed on ꢂ-Al2O3 after
ozonation by FTIR
The adsorption of the carboxylic acids on ␥-Al2O3 in presence of
ozone was also confirmed by FTIR (Fig. S7). In comparison to FTIR
spectrum of ␥-Al2O3 before ozonation several additional vibra-
tional bands of low intensity were found on the FTIR spectrum
of ␥-Al2O3 after ozonation after outgassing the sample at 150 ◦C
to remove the adsorbed water (large bending vibration of water
at 1639 cm−1): at 1254, 1312, 1375, 1392, 1460, 1505, 1550, 1572,
1589, 1615 cm−1. For each carboxylic acids four bands are expected:
the symmetric and antisymmetric modes of C=O streching are in
the ranges 1625–1687 and 1660–1740 cm−1, respectively, the C-
O-H in-plane bending at 1395–1440 cm−1 and the C-O stretch at
1315–1280 cm−1. However, no bands above 1625 cm−1 have been
observed in the spectrum of ␥-Al2O3 after ozonation revealing that
carboxylic acids are under their carboxylates form. Indeed, for car-
boxylate salts only two vibrations for each salts are expected as C=O
andC-O bonds of the acid are replaced by two equivalentC-O “bond-
Fig. 6. Adsorption isotherms of acetic acid (ꢃ), formic acid (×) and oxalic acid (ꢁ)
onto ␥-Al2O3 at 25 ◦C. [Acids]0: from 5 to 400 mg L−1
.
carboxylic will give a participation of 13% in the 46% TOC removal
with ␥-Al2O3, which is below to what was predicted (19%) consid-
ering 11% of direct ozonation reaction (as for single ozonation) and
16% of radical reaction. Three possibilities could be considered: (1)
other by-products could be adsorbed as dimethyl muconic acid,
(2) the contribution of direct ozonation is higher (17%) in presence
of alumina due to ozone adsorption at the surface of the particles
and (3) the amount of adsorbed carboxylic acids is higher in pres-
ence of ozone. Indeed during adsorption experiments of carboxylic
acids on ␥-Al2O3 the pH of the solution increased from 4.5 to 7.6.
In this range of pH carboxylic acids are under carboxylates form
(CH3COO−, HCOO−, −OOC-COO−) as pH is higher than the pKA of the
−
and-a half” bonds. These CO2 bonds interact out-of-phase and
in-phase to give two bands. The antisymmetric CO2 stretch band
is usually seen at 1540–1650 cm−1 and the symmetric CO2 stretch
band is usually seen at 1360–1450 cm−1. Acetate salts bands are
usually observed at 1400–1450 and 1550–1600 cm−1, formate at
1360 and 1600 cm−1 and oxalate at 1320 and 1620 cm−1. Consider-
ing the vibrational bands observed in the spectrum of ␥-Al2O3 after
ozonation, the bands relative to carboxylates adsorbed on ␥-Al2O3
surface could therefore be assigned as follows: for acetate bands at
1460 and 1550 cm−1; for formate bands at 1375 and 1589 cm−1
;
for oxalate bands at 1312 and 1615 cm−1. The remaining bands
correspond to unidentified adsorbed products (1254, 1392, 1505,
1572 cm−1), which could result from the adsorption of other car-
acids (pKA-CH
= 4.76, pKA-HCOOH = 3.74, pKA-HOOC-COOH = 1.25
COOH
3
and 4.27). The adsorption of the carboxylates (RCOO−) on ␥-Al2O3
occurs via a ligand exchange withAl-OH groups formingAl-O(CO)R
groups and liberates OH− responsible for the increase of pH. In
ozonation with ␥-Al2O3 the mechanism of adsorption is differ-
ent as in contrary a decrease of pH from 4.5 to 2.5 was observed.
At these pH carboxylic acids are under their protonated forms:
CH3COOH, HCOOH, HOOC-COO−, which could encountered for a
different mechanism of adsorption and maybe a higher adsorption.
Therefore, it is difficult from adsorption experiments to accurately
estimate the amount of carboxylic acids adsorbed on ␥-Al2O3 dur-
ing ozonation.
boxylate by-products as dimethyl muconic acid, as assumed by 13
C
NMR. Quantification of these products by FTIR was not possible as
in too low amounts.
3.7. Basic and acid sites of ꢂ-Al2O3 before and after ozonation by
FTIR
␥-Al2O3 is an amphoteric solid with Lewis acid AlOH(H+) sites
and basicAl-OH sites (Scheme 3). The acidic properties of ␥-Al2O3
before and after ozonation have been determined by FTIR after pyri-
dine adsorption on dried materials (Fig. S8). FTIR/pyridine spectrum
of ␥-Al2O3 before ozonation featured no band at 1547 cm−1 charac-
teristic of Brönsted acid site, and after ozonation a very weak band
appeared, which disappeared at 100 ◦C under vacuum, featuring
very low amount of weak Brönsted sites. This could come from free
COOH group on the surface of ␥-Al2O3 after ozonation as in the
case of oxalic acid adsorption. The band at 1596 cm−1 is relative
to H-bonding between OH surface groups and pyridine. This band
disappeared after outgassing at 100 ◦C. Three similar intense bands
at 1455, 1496 and 1625 cm−1 were observed in FTIR/pyridine spec-
tra of ␥-Al2O3 before and after ozonation, which remained after
outgassing at 500 ◦C. These bands are characteristic of strong Lewis
acid sites. The Lewis acid sites of ␥-Al2O3 were not modified dur-
ing ozonation. If the Lewis acid sites of ␥-Al2O3 participate to the
ozonation process they are entirely regenerated during ozonation
as already proposed in literature [8]. An example of simplified pos-
sible mechanism is given in Scheme 3. Furthermore, the acid sites
of ␥-Al2O3 are not affected by the adsorption of carboxylic acids,
they are not the adsorption sites.
3.6.2. Identification of carboxylic acids adsorbed on ꢂ-Al2O3 after
ozonation by 13C CP MAS NMR
In order to confirm the adsorption of carboxylic acids on ␥-Al2O3
in presence of ozone, solid state 13C CP MAS NMR was performed
on the materials after ozonation. 13C NMR spectrum (Fig. S6) exhib-
ited three major peaks with chemical shifts at 178, 164 and 21 ppm,
confirming the adsorption of acetic acid, formic acid and/or oxalic
acid at the surface of ␥-Al2O3 [46,47]. It is difficult to assign the
peak at 164 ppm to either formic acid or oxalic acid. Indeed in liq-
uid 13C NMR, formic acid is identified by a peak (COO) at 165.8 ppm
and oxalic acid by a peak (COO) at 160.9 ppm. For acetic acid the
two peaks at 178.1 (COO) and 22 (CH3) ppm were clearly identi-
fied. Additional peaks of lower intensity of unidentified products
are present on the spectrum at 142, 131, 81, 42 and 29 ppm. The
peaks at 142, 131 and 29 could come from the adsorption of some
dimethyl muconic acid, as muconic acid in solution presents three
peaks at 166, 140 and 129 ppm and the peak at 29 ppm could come
from the contribution of methyl groups. The peak at 42 ppm could
come from the adsorption of methyl malonic acid (Product 7 in
Scheme 2), as malonic acid in solution presents two peaks at 178
and 49 ppm. However, it is very difficult to quantify each of these
adsorbed products due to their very low amount on the material
The basic properties of ␥-Al2O3 before and after ozonation have
been determined by FTIR after CO2 adsorption on dried materi-
als (Fig. S9). For both materials, CO2 adsorption has revealed the
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
10
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
Scheme 3. Expected interaction of O3 with (a) acid and (b) basic sites of ␥-Al2O3 (scheme adapted and modified from Ref. [8]). (b) Hypothesis for adsorption mechanism of
carboxylic acids with the basic sites of ␥-Al2O3 during ozonation.
presence of Al3+ extra-framework species evidenced by a fine band
at 2365 cm−1 of CO2 is linearly coordinated to these Al species.
FTIR spectra of CO2 adsorbed on ␥-Al2O3 before and after ozona-
tion showed also similar four major peaks at 1231, 1482, 1644
and 1774 cm−1 characteristic of the formation of carbonates at
the surface of the material due to the chemisorption of CO2 on
the Al-OH basic sites. Lower amount of carbonates were observed
on ␥-Al2O3 after ozonation. Furthermore, chemisorbed carbonates
were desorbed at 250 ◦C for ␥-Al2O3 before ozonation featuring a
medium force of the basic sites and at lower temperature (200 ◦C)
on ␥-Al2O3 after ozonation revealing a slightly lower force of
the basic sites. Ozonation implied a decrease of the basicity of
Al-OH species in number and in force due most probably to car-
boxylates adsorption on the basic sites. An example of simplified
possible mechanisms is given in Scheme 3 to explain the com-
petition between Al-OH as active site for catalytic ozonation and
Al-OH as adsorption sites for carboxylates. This hypothesis allows
to explain the additional decrease of pH during ozonation with ␥-
Al2O3 due to the chemisorption of RCOOH on an intermediate site
Al(+) (formed during ozonation), which liberates a proton to give an
electrostatic interaction RCOO−–Al(+) and contributes to decrease
the pH (Scheme 3). The amount of Al-OH species could not be cal-
culated at 25 ◦C due to the presence of water hiding FTIR bands of
the OH region (3000–4000 cm−1). After outgassing the samples at
500 ◦C, which implies water removal but also carboxylates removal,
similar FTIR spectra of Al-OH bands at 3793, 3774, 3730, 3689,
3675 cm−1 were observed for ␥-Al2O3 before and after ozonation
with an amount of OH groups ∼1.1 mmol OH g−1 (Fig. S10). Basic
Al-OH sites are affected by carboxylates adsorption and their num-
ber decreased during ozonation, however the recovery of all Al-OH
groups after outgassing suggests a possible regeneration of ␥-Al2O3
by calcination at 500 ◦C.
ozonation allowed for the first run to reach a higher level of
TOC removal with 71% after 8 h with ␥-Al2O3, whereas for sin-
gle ozonation TOC removal was only 30% (Fig. 7). Then a series of
10 ozonation runs was performed using the same ␥-Al2O3 sam-
ple, in order to assess the stability of the material (Fig. 7). Each run
lasted about 8 h and ␥-Al2O3 was reused without any regenera-
tion or specific treatment. Millimetric ␥-Al2O3 particles were put
in a perforated basket easy to immerge in the solution and easy
to remove and to place in another batch of effluent. TOC removal
clearly progressively decreased during the successive runs with
␥-Al2O3. After 7–10 runs with ␥-Al2O3 TOC removal was quite
reproducible with 40% TOC removal, which remained higher than
single ozonation (30% TOC removal). The decrease of TOC removal
with ␥-Al2O3 is due to the decrease of the amount of basic Al-OH
sites by irreversible chemisorption of carboxylates (Scheme 3). If
for each run 0.15 mmol of carboxylates per gram of ␥-Al2O3 are
chemisorbed on Al-OH (as found in Section 3.6.1), seven runs will be
enough to eliminate all Al-OH groups of ␥-Al2O3 (1.1 mmol OH g−1),
which is consistent with the result obtained in this reuse study.
100
80
60
40
20
0
Cycle 1
Cycle 4
Cycle 7
Cycle 10
Cycle 2
Cycle 5
Cycle 8
O3
Cycle 3
Cycle 6
Cycle 9
0
60
120
180
240
Time (min)
300
360
420
480
3.8. Reuse and regeneration of ꢂ-Al2O3
3.8.1. Reuse of ꢂ-Al2O3 in successive ozonation reactions
Fig. 7. TOC removal during single ozonation (ꢂ) and during a series of 10 successive
ozone/␥-Al2O3 coupled treatment runs using the same solid without regenera-
tion. First run (᭹), fifth run (♦) and tenth run (ꢁ). Temperature: 25 ◦C; O3 flow
rate: 30 L h−1; [O3]Inlet: 5 g Nm−3; [2,4-DMP]0: 50 mg L−1; volume of solution: 2 L;
To establish the durability of ␥-Al2O3 during ozonation, the
material has been reused in 10 successive runs with a higher
amount of solid (5 g L−1) and a higher amount of entering O3
(5 g Nm−3 produced from O2). This slightly harder condition of
␥-Al2O3: 5 g L−1
.
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
11
100
80
60
40
20
0
250
200
150
100
50
0
60
120
180
Time (min)
240
300
360
420
480
0
8
0
100
200
300
Time (min)
Fig. 8. TOC removal during single ozonation (ꢂ) and during ozone/␥-Al2O3 coupled
treatment using a fresh sample of alumina (᭹) and a thermally regenerated sample
of alumina (ꢁ). Temperature: 25 ◦C; O3 flow rate: 30 L h−1; [O3]Inlet: 5 g Nm−3; [2,4-
DMP]0: 50 mg L−1; volume of solution: 2 L; ␥-Al2O3: 1.85 g L−1
.
The remaining stable activity of ␥-Al2O3 after 7–10 runs, which is
10% higher than single ozonation is due either to reversible cat-
alytic activity of Lewis sites of ␥-Al2O3 or to a higher efficiency of
direct O3 reaction at the solid surface. As for the previous study
16% TOC removal were assigned to radical reaction (Section 3.4),
radicals are produced either by acid sites or acidic sites (Scheme 3),
it seems that basic sites are also involved in the catalytic activity
of ␥-Al2O3. Basic sites of ␥-Al2O3 act in TOC removal as carboxylic
acids adsorbents and also as basic catalytic sites, which are lost at
each carboxylate adsorption.
6
4
2
0
100
200
300
t (min)
3.8.2. Regeneration of ꢂ-Al2O3 after ozonation
After 10 runs of ozonation (80 h ozonation), ␥-Al2O3 was
removed from the solution, dried at 120 ◦C during 20 min and
calcined at 500 ◦C during 6 h. A part of the resulting material
(1.85 g L−1) was then used in a new ozonation experiment and its
performance was compared to the same amount of a fresh ␥-Al2O3
sample. The thermally regenerated ␥-Al2O3 showed a similar effi-
ciency of the new ␥-Al2O3 with 60% TOC removal after 8 h (Fig. 8).
As observed earlier by FTIR (Section 3.7), a calcination at 500 ◦C
allowed to remove adsorbed carboxylates and to recover all basic
Al-OH groups of ␥-Al2O3. Therefore, a calcination at 500 ◦C allows
to regenerate and reuse this catalyst without any loss of perform-
ances.
Fig. 9. TOC and pH evolution during single ozonation (open symbols) and during
ozonation with ␥-Al2O3 (plain symbols) without (circles) and with (triangles) NaCl
50 g L−1 of a synthetic petrochemical effluent containing various petrochemicals:
200 mg L−1 phenol, 200 mg L−1 acetic acid, 25 mg L−1 naphtenic acid, 0.05 mg L−1
pyrene, 0.95 mg L−1 naphtalene. Temperature: 35 ◦C; O3 flow rate: 24 L h−1; [O3]Inlet
5 g Nm−3; [TOC]0: 250 mg L−1; volume of solution: 1.5 L; ␥-Al2O3: 2 g L−1
:
.
solid. A large amount of phenol (200 mg L−1, 2.1 mmol L−1) was
also added and if we consider a similar mechanism as for 2,4-
DMP (Section 3.5.4) 1 mol of phenol will produce 1.7 mol acetic
acid, 1 mol formic acid and 0.05 mol oxalic acid. This will corre-
spond to a production of 216 mg L−1 of acetic acid, 100 mg L−1
formic acid and 9 mg L−1 oxalic acid and therefore an adsorption
capacity (calculated from the adsorption isotherms in Fig. 6) on
␥-Al2O3 of 13 mg g−1 (0.22 mmol g−1), 10 mg g−1 (0.21 mmol g−1),
6 mg g−1 (0.07 mmol g−1) for each acid, respectively. The total of
expected carboxylates adsorbed on the solid will be therefore at
least 0.7 mmol g−1, which means that for only two molecules (acetic
acid and phenol) already ∼65% of Al-OH basic sites will be rapidly
blocked. The stopover of the reaction after 50 min with ␥-Al2O3
is therefore due to the total obstruction of Al-OH basic sites by
adsorbed carboxylates. For single ozonation a slightly higher TOC
removal is obtained after 300 min (58% TOC removal). The fact that
␥-Al2O3 presents no better activity than single ozonation when
all of the basic sites are blocked let suppose that the Lewis acid
sites of ␥-Al2O3 do not participate to the ozonation process. We
have previously observed that Lewis acid sites were not affected
during ozonation (Section 3.7). We therefore proposed that the cat-
alytic activity of oxides in water ozonation is due principally to the
basic sites (M-OH) of the materials (Scheme 3). The Lewis acid sites
previously demonstrated as responsible for catalytic ozonation in
3.9. Ozonation of a synthetic petroleum effluent
3.9.1. Ozonation of a synthetic petroleum effluent without any
additive
A synthetic petrochemical effluent containing various petro-
chemicals was treated with ␥-Al2O3. This experiment was
performed in a third pilot plant scale reactor described elsewhere
[28]. The complex synthetic effluent was a mixture of several
aromatic hydrocarbons and associated acids: phenol, acetic acid,
naphtenic acid, pyrene, naphtalene with initial concentration of
200, 200, 25, 0.05, 0.95 mg L−1, respectively, corresponding to
TOC = 230 mg L−1. With ␥-Al2O3 TOC removal was much faster
than for single ozonation during the first 50 min and then stopped
and remained at 50% equivalent to single ozonation (Fig. 9). In
the synthetic effluent a large amount (200 mg L−1) of acetic acid
was added, which blocked Al-OH basic sites of ␥-Al2O3 from the
beginning of the process. Considering the adsorption isotherm
(Fig. 6) a concentration of acetic acid of 200 mg L−1 will corre-
spond to 12 mg g−1 (0.20 mmol g−1) of acetate adsorbed on the
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
12
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
For pH < 3.3 Cl2 will be the main species, for 3.3 < pH < 7.5
HOCl and for pH > 7.5 OCl− [53]. At low initial pH (pH < 2.5) chlo-
rine species are able to degrade efficiently some carboxylic acids
(oxalic ꢀ citric > maleic acids) leading to a pH increase from 2.5 to
9 [53]. However, acetic acid is barely decomposed except if a new
source of HO• radical is produced as for instance by UV irradia-
tion [53]. However, under UV, generally an intermediate product
as H2O2 is produced as for O3/UV [50]:
gaseous phase [48,49] cannot participate to the reaction in water,
as these sites are surely blocked by water molecules that O3 cannot
displace.
3.9.2. Ozonation of a synthetic petroleum effluent in presence of
NaCl and toxicity tests
The effect of NaCl in ozonation processes is very important
as some wastewater contains some salts as NaCl, which can be
present in high concentration level in tannery and dye manu-
facturing wastewater, for instance, and also in lower amount in
produced water from off-shore oil extraction. An excess of NaCl
(50 g L−1) was added to the ozonation process whereas in seawater
NaCl concentration is around 10 g L−1. Adding NaCl to the solution
containing various chemicals (phenol, acetic acid, naphtenic acid,
pyrene, naphtalene) led to an increase in TOC removal for both pro-
cesses single ozonation and ozonation with ␥-Al2O3 (Fig. 9). The
inverse phenomenon was observed previously in ozonation of 2,4-
DMP with basic Na-LTA zeolites with the same amount of NaCl [18].
It was observed that Cl− ions were scavenging HO• radicals simi-
larly as t-BuOH for a pH solution varying from pH 5 to 7 and no effect
(or a slight negative effect) was observed for the single ozonation
[18]. It was showed in literature that the efficiency of chloride ions
as HO• scavenger increases by a factor of 100 then pH decreases
from 6 to 3 and then the concentration of NaCl increases from 1.5
to 73 g L−1 [50]. For low salinity water (<1.5 g L−1) and at pH 5 no
effect of chloride ions was observed. For concentration of NaCl > 75,
50 and 10 g L−1 75, 70 and 25% of HO• were removed, respectively.
O3 + H2O → H2O2
UVlightat254 nm(lowpressurelamp)
O3 + H2O2 → 2HO• + 3O2
As in our study HO• are most probably highly scavenged by
Cl−, therefore we propose that H2O2 could be also produced as
an intermediary product during ozonation of PAH in presence of
NaCl to explain the higher oxidation level of single ozonation and
the mineralization of acetic acid. Indeed H2O2 could react with
acetic acid to give peracetic acid, which is a powerful oxidant.
We suggest that PAH compounds react with Cl• or Cl2 to pro-
duce H2O2 and HO•. Indeed PAHs are known to decompose in
water by chlorine [55] and give the corresponding quinones [56].
Industrially H2O2 is produced from the oxidation of dihydroxyan-
thracene by O2 giving anthraquinone and H2O2. For ozonation in
presence of NaCl and PAH we suggest the formation of an inter-
mediate dihydroxypolyaromatic compounds giving corresponding
quinone and H2O2. It has been demonstrated that pyrene with
chlorine decomposes into pyrene-4,5-dione and 1-chloropyrene
by both oxidation and chlorine-substitution reactions, respectively
[57]. Authors evidenced that pyrene decomposition was associ-
ated with the formation of other reactive species than chlorine and
they suggest the formation of HO• radicals. These reactive species
produced during pyrene decomposition in chlorination have not
been identified but are reacting with ethanol. As we previously
showed that acetic acid could not decomposed by HO• radicals,
we suggest that one of these reactive species could be H2O2. The
increase of TOC removal in the presence of NaCl was accompa-
nied by the increase of the pH of the solution (Fig. 9) from 3.5
to 6 with ␥-Al2O3 after 25 min revealing the degradation of the
sodium carboxylates liberating CO2 and HO−. For single ozona-
tion, the pH of the solution increased from 3 to 7 after 100 min
due to the degradation of the carboxylic acids. The faster efficiency
observed with ␥-Al2O3 is most probably due to the faster forma-
tion of Cl• from HO• in comparison to the slow formation of Cl2
from O3. After 5 h the two processes have equivalent TOC removal
(90%).
It was previously demonstrated in literature that the toxicity
of a solution containing pollutants increased when single ozona-
tion was performed in presence of NaCl [54]. Toxicity tests were
therefore performed for ozonation with ␥-Al2O3 in presence of
NaCl using two non-standardized toxkits named ARTOXKIT and
ROTOXKIT exploiting two different seawater organisms Artemia
franciscana and Brachionus calyciflorus, respectively. The toxicity
tests were performed after pH adjustment at pH 7 of the syn-
thetic effluent treated after 150 and 300 min of ozonation with
␥-Al2O3. The efficiency of the process was assessed by the efflu-
ent concentration able to induce 50% of mortality of the species
after 24 h of exposal to the aqueous solution (LC50 24 h). The results
were expressed in toxic unit: Equitox m−3 = 100/LC50 24 h (Table 4).
After 150 min of ozonation with ␥-Al2O3 the toxicity level of the
solution increased revealing that the first by-products produced
during ozonation are more toxic than the initial pollutants. After
300 min of ozonation with ␥-Al2O3 the toxicity level of the solution
decreased. The synthetic effluent, which was initially considered as
slightly toxic, became non-toxic for at least one seawater organism.
As in the present study the salinity of the solution is high (50 g L−1
)
and the pH of the solution is acidic (4 > pH > 3) chloride ions should
present a very strong scavenging effect on HO• radicals produced
during ozonation with ␥-Al2O3. The equations corresponding to the
reaction of chloride ions with HO• are the following [50]:
HO• + Cl− → HOCl−•
HOCl−• + H+ → Cl• + H2O
pK = 7.2
The first reaction is fast and the second reaction is five times
more rapid than the first one. At pH < 7.2 Cl• is the dominant species.
Although HO• should be seriously scavenged by Cl− a remarkable
positive effect of NaCl was noticed with TOC removal reaching 90%
against 50% without NaCl. This high level of TOC removal indicates
that acetic acid has been mineralized during the ozonation process
with NaCl. NaCl avoid carboxylates to block the basic sites of ␥-
Al2O3 with presumably sodium cations interacting with carboxylic
acids to give the corresponding salts. Considering single ozona-
tion the only difference with our previous study about ozonation in
presence of zeolites and NaCl [18] is the solution composition and
mainly the presence of polycyclic aromatic hydrocarbons (PAH).
Ozone in presence of chloride anions in acidic aqueous medium
reacts following the proposed equations [51,52]:
O3 + Cl− → O2 + ClO−
(ratedeterminingreaction, veryslowreaction)
2H+ + Cl− + ClO− → Cl2 + H2O
(rapid)
Cl2 has been observed as the main and only product resulting
from O3 decomposition in presence of NaCl [52] although Cl2 in
water is in equilibrium with other products [53]:
Cl2 + H2O → HOCl + H+ + Cl−
pK1 = 3.3
HOCl → OCl− + H+
pK2 = 7.5
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
13
Table 4
Percentage of TOC removal and toxicity index of the synthetic petrochemical effluent (phenol, acetic acid, naphthenic acid, pyrene, naphtalene) after ozonation with ␥-Al2O3
after 150 and 300 min treatment.
TOC removal (%)
ARTOXKIT (toxic unita)
ROTOXKIT (toxic unita)
Toxicity
Initial solution
O3/␥-Al2O3 (150 min)
O3/␥-Al2O3 (300 min)
0
82
89
5.2
10.1
2.6
5.9
8.4
5.0
Slightly toxic
Slightly toxic
Non-toxic and slightly toxic
a
Toxic unit: non-toxic < 3; 3 < slightly toxic < 10; 10 < toxic < 50; 50 < very toxic < 100; extremely toxic > 100. Temperature: 35 ◦C; O3 flow rate: 24 L h−1; [O3]Inlet: 5 g Nm−3
;
TOCInitial = 230 mg L−1; volume of solution: 1.5 L; ␥-Al2O3: 2 g L−1; NaCl: 50 g L−1
.
Catalytic ozonation in presence of NaCl with ␥-Al2O3 allowed to
decrease the toxicity of the petrochemical effluent.
1.5 h) and then to a decrease of the toxicity below the initial tox-
icity (after 5 h). ␥-Al2O3 is therefore a good candidate for catalytic
ozonation of refinery wastewater containing refractory PAH and
should be better performing in lower salinity water as seawater.
The large controversy observed in ozonation in water is because
ozonation depends of a lot of parameters as pH, presence of salts,
nature of the catalysts, but also is effluent composition dependent.
4. Conclusion
The effect of ␥-Al2O3 during ozonation for 2,4-DMP removal
was investigated through different experiments. It appeared clearly
that the material allowed increasing the oxidation efficiency of the
initial pollutant in comparison to single ozonation. The beneficial
effect of ␥-Al2O3 was linked to two phenomena: its adsorption
capacity for carboxylic acids by-products (acetic acid, formic acid
and oxalic acid) and its ability to generate •OH radicals. The
identification of different by-products during 2,4-DMP ozonation
confirmed that both mechanisms of direct O3 reaction and radical
type reaction coexisted during ozonation with ␥-Al2O3, in contrary
to single ozonation, which is mainly governed by direct O3 reaction.
Radical type reaction for ozonation with ␥-Al2O3 produced initially
more by-products (other than acetic, formic and oxalic acids) than
single ozonation leading at the end to less carboxylic acids for-
mation. After 5 h ozonation similar amount of by-products (other
than carboxylic acids) were found for both processes, but final by-
products were more oxidized with ␥-Al2O3 than those resulting of
single ozonation. Cumulative adsorption of carboxylic acids on the
basic Al-OH sites of ␥-Al2O3 resulted in a progressive decrease of
the TOC removal during successive runs of ozonation until all Al-OH
groups were blocked by carboxylates. Lewis acid sites of ␥-Al2O3
were not affected during ozonation. A complete regeneration of Al-
OH basic sites was obtained by thermal treatment at 500 ◦C without
altering the properties of the solid. Finally, ␥-Al2O3 was used to
treat a synthetic petrochemical effluent containing different pol-
lutants (phenol, naphtenic acid, pyrene, naphtalene) and a large
amount of acetic acid. TOC removal with ␥-Al2O3 was very fast in
the first hour of ozonation, much faster than single ozonation, and
then stopped due to the inhibition of all Al-OH groups by carboxyl-
ates. A similar activity as single ozonation was then obtained with
50% TOC removal. This result allows us to propose that only Al-OH
basic sites are involved in catalytic ozonation in water. By adding
NaCl to the ozonation processes remarkable higher TOC removals
were reached for single ozonation and for ozonation with ␥-Al2O3
with 90% TOC after 5 h. Two antagonism mechanisms occur with
␥-Al2O3: sodium cations prevent the adsorption of carboxylates on
the basic sites of ␥-Al2O3 by forming preferentially sodium carbox-
ylates in solution, while chloride ions are strong •OH scavengers
especially in the conditions of the study (low pH and high NaCl
concentration). The higher TOC removal is concomitant with the
increase of pH due presumably to the degradation of carboxylates
liberating hydroxyl ions. The reaction is faster with ␥-Al2O3 in com-
parison to single ozonation due to different mechanism pathways
between chloride ions with •OH and with O3, respectively. The
enhancement of the mineralization of the solution is also conjointly
most probably induced by the formation of H2O2 resulting from oxi-
dation reactions between polycyclic aromatic hydrocarbon (PAH)
and •Cl or Cl2 for ozonation with ␥-Al2O3 and for single ozonation,
respectively. Ozonation with ␥-Al2O3 in presence of large amount
of NaCl led to a first increase of the toxicity of the effluent (after
Acknowledgments
The authors acknowledge the French National Agency for
Research (ANR) for supporting this study through the convention
ANR ECOTECH 2010 project PETZECO (1081C0230/ANR-10-ECOT-
011-03). Special thanks are addressed to Bruno Navarra for his help
for the development of the ozonation pilot, to Eddy Petit (HPLC-
UV), Valérie Bonniol (Ionic chromatography) and Guillaume Cazals
(LC-MS) for their precious help in by-products identification.
References
[1] M. Petala, V. Tsiridis, P. Samaras, A. Zouboulis, G.P. Sakellaropoulos, Desalina-
tion 239 (2009) 229–246.
[2] P. Pocostales, P. Alvarez, F.J. Beltran, Chem. Eng. J. 168 (2011) 1289–1295.
[3] B. Kasprzyk-Hordern, P. Andrzejewski, A. Dabrowska, K. Czaczyk, J. Nawrocki,
Appl. Catal. B:Environ. 51 (2004) 51–66.
[4] F.J. Beltran, F.J. Rivas, R. Montero-de-Espinosa, Appl. Catal. B: Environ. 47 (2004)
101–109.
[5] J. Reungoat, J.S. Pic, M.H. Manero, H. Debellefontaine, Separ. Sci. Technol. 42
(2007) 1447–1463.
[6] T. Merle, J.S. Pic, M.H. Manero, S. Mathe, H. Debellefontaine, Catal. Today 151
(2010) 166–172.
[7] J. Reungoat, J.S. Pic, M.H. Manero, H. Debellefontaine, Water Sci. Technol. 62
(2010) 1076–1083.
[8] J. Nawrocki, B. Kassprzyk-Hordern, Appl. Catal. B: Environ. 99 (2010) 27–42.
[9] C. Cooper, R. Burch, Water Res. 33 (1999) 3689–3694.
[10] H. Fujita, J. Izumi, M. Sagehashi, T. Fujii, A. Sakoda, Water Res. 38 (2004)
166–172.
[11] Y. Dong, H. Yang, K. He, X. Wu, A. Zhang, Appl. Catal. B: Environ. 82 (2008)
163–168.
[12] T. Merle, J.S. Pic, M.H. Manéro, S. Mathé, H. Debellefontaine, Ozone-Sci. Eng. 32
(2010) 391–398.
[13] J. Reungoat, J.S. Pic, M.H. Manéro, H. Debellefontaine, Water Sci. Technol. 62
(2010) 1076–1083.
[14] H. Valdés, S. Alejandro, C.A. Zaror, J. Hazard. Mater. 227 (2012) 34–40.
[15] L. Qi, H. You, Z. Zhang, C. Feng, S. van Agtmaal, Int. J. Electrochem. Sc. 8 (2013)
5457–5468.
[16] F. Qi, B. Xu, Z. Chen, L. Feng, L. Zhang, D. Sun, Chem. Eng. J. 219 (2013) 527–536.
[17] W. Aboussaoud, M.H. Manéro, J.S. Pic, H. Debellefontaine, Ozone: Sci. Eng
(2014), accepted manuscript.
[18] J. Vittenet, J. Rodriguez, E. Petit, D. Cot, J. Mendret, A. Galarneau, S. Brosillon,
Microporous Mesoporous Mater. 189 (2014) 200–209.
[19] H. Fujita, K. Shiraishi, T. Fujii, A. Sakoda, J. Izumi, Adsorption 11 (2005) 835–839.
[20] A. Ikhlaq, D.R. Brown, B. Kasprzyk-Hordern, Appl. Catal. B: Environ. 129 (2013)
437–449.
[21] N. Brodu, M.H. Manéro, C. Andriantsiferana, J.S. Pic, H. Valdés, Chem. Eng. J. 231
(2013) 281–286.
[22] M. Ernst, F. Lurot, J.-C. Schrotter, Appl. Catal. B: Environ. 47 (2004) 15–25.
[23] T. Zhang, C. Li, J. Ma, H. Tian, Z. Qiang, Appl. Catal. B: Environ. 82 (2008) 131–137.
[24] F. Qi, Z. Chen, B. Xu, J. Shen, J. Ma, C. Joll, A. Heitz, Appl. Catal. B: Environ. 84
(2008) 684–690.
[25] US Environmental Protection Agency, Ambient Water Criteria for 2,4-
Dimethylphenol, Environmental Protection of Water Regulations and Stan-
dards Division, Washington, DC, 1990.
[26] J.P. Ghosh, K.E. Taylor, J.K. Bewtra, N. Biswas, Chemosphere 71 (2008)
1709–1717.
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037

G Model
APCATA-15069; No. of Pages14
ARTICLE IN PRESS
14
J. Vittenet et al. / Applied Catalysis A: General xxx (2014) xxx–xxx
[27] A. Galarneau, D. Desplantier, R. Dutartre, F. Di Renzo, Microporous Mesoporous
Mater. 27 (1999) 297–308.
[28] J. Bergraser, J. Savignac, M. Jacob, P. Pedenaud, N. Lesage, IOA Proceedings,
Berlin, 2013.
[29] M. Akia, S.M. Alavi, Z.F. Yan, Petrol. Coal 52 (2010) 208–289.
[30] M. Trapido, Y. Veressinina, R. Munter, J. Environ. 124 (1998) 690–694.
[31] L. Zhao, J. Ma, Z. Sun, Appl. Catal. B: Environ. 79 (2007) 244–253.
[32] A. Kahru, L. Pollumaa, Oil Shale 23 (2006) 53–93.
[33] E.B. Azevedo, F.R. de Aquino Neto, M. Dezotti, J. Hazard. Mater. 128 (2006)
182–191.
[44] C.-R. Huang, H.-Y. Shu, J. Hazard. Mater. 41 (1995) 47–64.
[45] Q. Liu, B. Shentu, C. Gu, J. Zhu, Z. Weng, Aiche J. 55 (2009) 2716–2724.
[46] J.B. Stothers, Carbon-13 NMR Spectroscopy, Academic Press, 1972.
[47] H.O. Kalinowsky, S. Berger, S. Braun, 13C NMR Spectroskopie, Georg Thieme
Verlag, 1984.
[48] K.M. Bulanin, J.C. Lavalley, A.A. Tsyganenko, J. Phys. Chem. 99 (1995)
10294–10298.
[49] K.M. Bulanin, J.C. Lavalley, A.A. Tsyganenko, Coll. Surf. A: Phys. Eng. Aspects 101
(1995) 153–158.
[50] C.-H. Liao, S.-F. Kang, F.-A. Wu, Chemosphere 44 (2001) 1193–1200.
[51] L.R.B. Yeatts, H. Taube, J. Amer. Chem. Soc. 71 (1949) 4100–4105.
[52] A.V. Levanov, I.V. Kuskov, E.E. Antipenko, V.V. Lunin, Russian J. Phys. Chem. A
(2012) 757–762.
[34] J.P. Scott, D.F. Ollis, Environ. Progr. 14 (2005) 88–103.
[35] F. Fourcade, S. Yahiat, K. Elandaloussi, S. Brosillon, A. Amrane, Chem. Eng. Tech-
nol. 35 (2012) 1–11.
[36] M. Gurol, P. Singer, Water Res. 3 (1983) 1173–1181.
[37] H. Sang-Kuk, I. Kazuhiro, H. Utsumi, Water Res. 32 (1998) 3261–3266.
[38] T. Poznyak, R. Tapia, J. Vivero, I. Chairez, J. Mex. Chem. Soc. 50 (2006) 28–35.
[39] N.A.S. Amin, J. Akhtar, H.K. Rai, Chem. Eng. J. 158 (2010) 520–527.
[40] Y. Lui, J. Shen, Z. Chen, L. Yang, Y. Han, Appl. Catal. A: Gen. 403 (2011) 112–118.
[41] E. Mvula, C. Von Sonntag, Org. Biomol. Chem. 1 (2003) 1749–1756.
[42] M.K. Ramseier, U.V. Gunten, Ozone-Sci. Eng. 31 (2009) 201–215.
[43] T.N. Das, J. Phys. Org. Chem. 22 (2009) 872–882.
[53] Y.-J. Shih, K.-H. Chen, Y.-H. Huang, J. Taiwan Inst. Chem. Engineers 45 (2014)
962–966.
[54] E.B. Azevedo, F.R.D. Neto, M. Dezotti, J. Hazard. Mater. 128 (2006) 182–191.
[55] C. Rav-Acha, R. Blits, Water Res. 19 (1985) 1273–1281.
[56] J. Tillner, C. Hollard, C. Bach, C. Rosin, F.-F. Munoz, X. Dauchy, J. Chrom. A 1315
(2013) 36–46.
[57] K. Kosaka, H. Koshino, N. Yoshida, S. Kunikane, J. Water Supply Res.
Technol.—Aqua 59 (2010) 41–52.
Please cite this article in press as: J. Vittenet, et al., Appl. Catal. A: Gen. (2014), http://dx.doi.org/10.1016/j.apcata.2014.10.037