Sonochemical degradation of ethyl paraben in environmental samples: Statistically important parameters determining kinetics, by-products and pathways

Article abstract of DOI:10.1016/j.ultsonch.2015.12.002

The sonochemical degradation of ethyl paraben (EP), a representative of the parabens family, was investigated. Experiments were conducted at constant ultrasound frequency of 20 kHz and liquid bulk temperature of 30 °C in the following range of experimental conditions: EP concentration 250-1250 μg/L, ultrasound (US) density 20-60 W/L, reaction time up to 120 min, initial pH 3-8 and sodium persulfate 0-100 mg/L, either in ultrapure water or secondary treated wastewater. A factorial design methodology was adopted to elucidate the statistically important effects and their interactions and a full empirical model comprising seventeen terms was originally developed. Omitting several terms of lower significance, a reduced model that can reliably simulate the process was finally proposed; this includes EP concentration, reaction time, power density and initial pH, as well as the interactions (EP concentration) × (US density), (EP concentration) × (pH_o) and (EP concentration) × (time). Experiments at an increased EP concentration of 3.5 mg/L were also performed to identify degradation by-products. LC-TOF-MS analysis revealed that EP sonochemical degradation occurs through dealkylation of the ethyl chain to form methyl paraben, while successive hydroxylation of the aromatic ring yields 4-hydroxybenzoic, 2,4-dihydroxybenzoic and 3,4-dihydroxybenzoic acids. By-products are less toxic to bacterium V. fischeri than the parent compound.

Full text of DOI:10.1016/j.ultsonch.2015.12.002

Ultrasonics Sonochemistry 31 (2016) 62–70
Contents lists available at ScienceDirect
Ultrasonics Sonochemistry
journal homepage: www.elsevier.com/locate/ultson
Sonochemical degradation of ethyl paraben in environmental samples:
Statistically important parameters determining kinetics, by-products
and pathways
Costas Papadopoulos a, Zacharias Frontistis a, Maria Antonopoulou b, Danae Venieri c,
Ioannis Konstantinou b, Dionissios Mantzavinos a,
⇑
a Department of Chemical Engineering, University of Patras, Caratheodory 1, University Campus, GR-26504 Patras, Greece
b Department of Environmental & Natural Resources Management, University of Patras, 2 Seferi St., GR-30100 Agrinio, Greece
c School of Environmental Engineering, Technical University of Crete, Polytechneioupolis, GR-73100 Chania, Greece
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 28 August 2015
Received in revised form 3 November 2015
Accepted 4 December 2015
Available online 8 December 2015
The sonochemical degradation of ethyl paraben (EP), a representative of the parabens family, was inves-
tigated. Experiments were conducted at constant ultrasound frequency of 20 kHz and liquid bulk temper-
ature of 30 °C in the following range of experimental conditions: EP concentration 250–1250 lg/L,
ultrasound (US) density 20–60 W/L, reaction time up to 120 min, initial pH 3–8 and sodium persulfate
0–100 mg/L, either in ultrapure water or secondary treated wastewater.
A factorial design methodology was adopted to elucidate the statistically important effects and their
Keywords:
interactions and
a full empirical model comprising seventeen terms was originally developed.
Sonodegradation
Factorial design
Intermediates
Micro-pollutants
Aqueous matrix
Estrogenicity
Omitting several terms of lower significance, a reduced model that can reliably simulate the process
was finally proposed; this includes EP concentration, reaction time, power density and initial pH, as well
as the interactions (EP concentration) ꢀ (US density), (EP concentration) ꢀ (pHo) and (EP concentra-
tion) ꢀ (time).
Experiments at an increased EP concentration of 3.5 mg/L were also performed to identify degradation
by-products. LC–TOF–MS analysis revealed that EP sonochemical degradation occurs through dealkyla-
tion of the ethyl chain to form methyl paraben, while successive hydroxylation of the aromatic ring yields
4-hydroxybenzoic, 2,4-dihydroxybenzoic and 3,4-dihydroxybenzoic acids. By-products are less toxic to
bacterium V. fischeri than the parent compound.
Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction
concentrations at the ng/L level, although in certain cases these
reached the lg/L level [5]. Parabens are not adsorbed in the sludge
Parabens, esters of 4-hydroxybenzoic acid with an alkyl (from
ethyl to butyl) or benzyl group, have been employed for about a
century as preservatives in foodstuff, cosmetics and pharmaceuti-
cals and personal care products [1]. Several studies published at
the turn of the millennium suggested possible estrogenic activity
[2,3] and carcinogenic potential [4]. In this respect and although
parabens were generally considered harmless to human beings
for a long time, several concerns have been raised over the past
twenty years about parabens safety with emphasis given on their
endocrine disrupting potential [5].
and mostly remain in the liquid phase where they can be degraded
relatively easily but not completely [6]; therefore, WWTPs are con-
sidered the major point source of parabens in the environment due
to their incomplete removal [5]. Consequently, parabens re-enter
the aquatic environment and they are typically found at the ng/L
level [1].
Unlike other emerging micro-pollutants, the advanced oxida-
tion of parabens has received considerably less attention possibly
due to the facts that (i) suitable detection techniques were only
developed in the past 10–15 years [1], and (ii) their adverse effects
to living organisms are still arguable. Steter et al. [7] studied the
electrochemical oxidation of methyl paraben on boron-doped dia-
mond, while Hernandez-Leal et al. [8] studied the removal of four
parabens with C1–C4 alkyl groups by ozonation or activated
carbon adsorption in various water matrices. Dobrin et al. [9]
Monitoring campaigns in wastewater treatment plants
(WWTPs) in Europe, North America and Japan showed influent
⇑
Corresponding author.
E-mail address: mantzavinos@chemeng.upatras.gr (D. Mantzavinos).
http://dx.doi.org/10.1016/j.ultsonch.2015.12.002
1350-4177/Ó 2015 Elsevier B.V. All rights reserved.

C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
63
proposed a hybrid system coupling non-thermal plasma and
ozonation to degrade methyl paraben and concluded that the inte-
grated process was more effective than the individual ones in
terms of mineralization. Furthermore, several photochemical [10]
and TiO2-based photocatalytic processes [11–13] have also been
tested to degrade various parabens.
achieved through a photodiode array detector (Waters 2996 PDA
detector, detection k = 254 nm).
The evolution of sulfate ions during the process was followed by
a Dionex ICS-1500 instrument equipped with an ASRS Ultra II con-
ductivity detector and IonPac AS9-HC anionic column. The mobile
phase was an aqueous sodium carbonate (9 mM) solution at a flow
In recent years, ultrasound irradiation has been widely
employed to degrade successfully various micro-pollutants includ-
ing endocrine disrupting compounds (EDCs) such as bisphenol A
[14,15], estrogens [16,17], 4-cumylphenol [18] and phthalates
[19]. Nevertheless, the literature on the sonochemical degradation
of parabens is very limited consisting only of a couple of recent
studies. Sasi et al. [20] investigated the high frequency (200–
1000 kHz) sonochemical degradation of methyl paraben with
emphasis on the effect of operating variables and the water matrix,
while Daghrir et al. [21] reported that the sonochemical degrada-
tion of butyl paraben at 518 kHz was enhanced when the system
was simultaneously irradiated by UV-C light.
The purpose of this work was to study the low frequency
(20 kHz) sonochemical degradation of ethyl paraben (EP) and eval-
uate the statistically significant parameters that may determine
degradation rates implementing a factorial design methodology.
Six parameters were tested, namely EP concentration, ultrasound
power, reaction time, water matrix, initial pH and the addition of
sodium persulfate. Early-stage transformation by-products were
also identified and a plausible reaction network was suggested.
rate of 1 mL/min, while the injection volume was 25 lL.
LC–TOF–MS (liquid chromatography-time of flight mass spec-
trometry) system was applied for the identification of transforma-
tion by-products (TBPs) of EP. Prior to analysis, 2 mL of treated
solutions were extracted by means of a solid-phase extraction
(SPE), reported in our previous work [12], using Oasis HLB (divinyl
benzene/N-vinylpyrrolidone copolymer) cartridges (60 mg, 3 mL)
from Waters (Mildford, MA, USA). The LC system consisted of an
Ultra-High Performance LC pump (Dionex Ultimate 3000, Thermo)
incorporating a column thermostat and an autosampler interfaced
to a Focus microTOF II – time of flight mass spectrometer (Brüker
Daltonics, Germany). The MS part was operated using microTOF
control (version 2.0) software. The scan range applied in the full-
scan mode was m/z 50–500 at a scan rate 1 Hz. The chromato-
TM
graphic separations were run on
100 mm ꢀ 2.1 mm, 2.2
San Jose, USA) at 30 °C. The injected sample volume was 10
a
C18 Acclaim RSLC,
l
m particle size (Thermo Fisher Scientific,
L.
l
Mobile phases A and B were water with 0.1% formic acid and ace-
tonitrile, respectively at a flow-rate of 0.2 mL/min. Analysis was
performed by ESI source in negative ionization mode. A linear gra-
dient progressed from 5% B (initial conditions) to 99.9% A in 12 min
(maintained for 2 min), returned to the initial conditions after
1 min and finally re-equilibration time was set at 3 min. The ESI-
source parameters were as follows: dry gas flow rate 8 L/min
(nitrogen), nebulizer pressure 2.0 bar, capillary voltage at 3200 V,
end plate offset at 500 V, collision cell RF 70.0 Vpp, dry tempera-
ture at 220 °C. Prior to analysis, the TOF mass analyzer was exter-
nally calibrated using sodium formate, in the scan range m/z 50–
1000, to ensure mass accuracy with 5 ppm. Data were acquired
with the HyStar 3.2 software and analysed with Data Analysis 4.1
software package. In addition, chemical formula calculator,
included in Data Analysis software was used to provide chemical
formula and mass accuracy values. The identification of the major-
ity of the TBPs was also verified by comparison of retention time,
high resolution mass and MS spectra to the commercially available
standards.
2. Materials and methods
2.1. Materials
Ethyl paraben (EP) (HO–C6H4–CO–O–CH2CH3, CAS no: 120-47-
8) and sodium persulfate (SPS) (Na2S2O8, 99+%, CAS number
7775-27-1) were supplied by Sigma–Aldrich and used as received.
Two water matrices were employed, i.e. ultrapure water (UPW,
pH = 6.5) taken from a water purification system (EASYpureRF-Ba
rnstead/Thermolyne, USA), and secondary treated wastewater
(WW) taken from the university campus treatment plant (pH = 8,
COD = 21 mg/L). Sulfuric acid or sodium hydroxide was used, as
needed, to adjust the initial solution pH of about 6 to acidic or alka-
line conditions.
2.2. Ultrasound irradiation
2.4. Acute ecotoxicity
A Branson 450 horn-type digital sonifier operating at a fixed fre-
quency of 20 kHz was employed. Reactions took place in a cylindri-
cal, double-walled, Pyrex vessel, which was open to the
atmosphere. Ultrasound irradiation was emitted through a 1 cm
in diameter titanium tip which was positioned in the middle of
the vessel at a distance of 3 cm from the bottom. The working vol-
ume was 0.12 L and the bulk temperature was kept constant at
30 °C with a temperature control unit. The maximum nominal
power output of the sonifier was 450 W and the actual energy
transmitted to the liquid phase was determined calorimetrically;
experiments were performed at actual power densities of 20 and
60 W/L.
The marine bacterium Vibrio fischeri was used to assess the
acute ecotoxicity of EP prior to and after sonodegradation. Changes
in bioluminescence of V. fischeri exposed to EP solutions for 15 min
were measured using a LUMIStox analyzer (Dr. Lange, Germany)
and the results were compared to an aqueous control.
2.5. Yeast estrogen screening (YES)
The YES assay using the yeast Saccharomyces cerevisiae was car-
ried out to assess the estrogenicity of EP according to the proce-
dures described in detail elsewhere [22]. In brief, standard
solutions and sample extracts were produced in ethanol and
2.3. Chromatographic techniques
10 lL of dilution series were dispensed into triplicate wells of
96-well microtiter plates. After evaporation to dryness at room
temperature, 0.2 mL of growth medium containing the chro-
mogenic substrate chlorophenol red-b-D-galactopyranoside
(CPRG) and the yeast cells were added, followed by incubation at
32 °C for 72 h. Each plate contained at least one row of blanks
and a standard curve for 17b-estradiol (E2). During the incubation
period, the microtiter plates were shaken at 80 rpm for 2 min to
mix and disperse the growing cells. The absorbance of the medium
High performance liquid chromatography (HPLC: Alliance 2695,
Waters) was employed to monitor the concentration of EP. Separa-
tion was achieved on a Kinetex XB-C18 100A column (2.6
lm,
2.1 mm ꢀ 50 mm) and a 0.5
l
m inline filter (KrudKatcher Ultra)
both purchased from Phenomenex. The mobile phase consisting
of 75:25 water:acetonitrile eluted isocratically at 0.35 mL/min
and 45 °C, while the injection volume was 40 lL. Detection was

64
C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
was measured using a microplate reader (LT-4000MS Microplate
Reader, Labtech) and Manta PC analysis software. The absorbance
at 540 nm was regarded as estrogenic activity after subtraction
of absorbance at 620 nm to correct for yeast growth. Positive wells
are indicated by a deep red color and by turbid yeast growth.
Equally important is EP concentration since the concentration
ratio of continuously sono-generated ROS to batch-fed EP will
determine degradation kinetics and apparent orders of reaction
(as will be discussed in Section 3.3). Since emerging micro-
pollutants are likely to appear in various water matrices including
groundwater and WWTP effluents, the effect of matrix complexity
(e.g. organic and inorganic constituents of WW) on degradation
should be assessed in comparison to e.g. UPW.
3. Results and discussion
Fig. 2 shows typical results concerning EP sonochemical degra-
dation as a function of six parameters, namely: reaction time,
ultrasound power density, EP concentration, water matrix and
pH, and the presence of SPS. Although one could guess correctly
that the amount of EP degraded depends on its initial concentra-
tion and the reaction time, the effect of other parameters may
not be evident.
3.1. How estrogenic is ethyl paraben?
Although parabens have been suspected for endocrine disrupt-
ing behavior, information regarding their estrogenicity is scarce
[2]. In view of this, the YES assay was employed to evaluate EP
estrogenicity relative to 17b-estradiol (E2) and the results are
shown in Fig. 1. EP is slightly estrogenic in the range of concentra-
In this work, a statistical approach was chosen based on a facto-
rial experimental design that would allow us to infer about the
individual and combined effects of the parameters with a relatively
small number of experiments. The independent parameters of the
experimental design are presented in Table 1. Each one of the six
variables received two values, a high value (indicated by the plus
sign) and a low value (indicated by the minus sign).
tions tested (i.e. 250 lg/L–8 mg/L including the concentrations of
the factorial design as will be discussed in Section 3.2); in fact,
its estrogenic activity at the maximum concentration of 8 mg/L is
about 4 times lower than that of reference E2 at its lowest concen-
tration of 62.5 lg/L. The inset of Fig. 1 shows a pictorial represen-
tation of the assay, where changes from yellow to brown to red
color are characteristic of a transition from zero to moderate to
intense estrogenicity. Routledge et al. [2] reported that parabens,
including EP, were several times less responsive to the YES test
than E2.
EP concentration levels between the very low mg/L and low
lg/L were chosen in this work, which (i) allows the accurate
quantitation of residual EP down to few g/L with the analytical
l
techniques available in this work, and (ii) covers, to a certain
degree, the environmentally relevant concentrations of parabens
that take values between ng/L and lg/L [1,5].
3.2. Selection of parameters and factorial design
The lower power density was chosen to secure a sufficient
degradation rate since preliminary experiments showed little
sonochemical activity below 20 W/L, while the upper power den-
sity of 60 W/L was selected to allow for the fast removal of dissi-
pated heat, thus keeping the liquid bulk temperature constant.
The addition of SPS at 100 mg/L was based on the findings of a
recent study of our group [15] regarding the sonochemical degra-
dation of endocrine disruptor bisphenol A at similar operating
conditions.
The initial pH was varied between 3 and 8, whereas the lower
pH values stand for wastewaters from the cosmetics/pharmaceuti-
cals industry, while the upper value matches that of the municipal
WW sample. Although the solution was not buffered to the afore-
mentioned pH values, pH was monitored constantly throughout
the reaction showing that only marginal changes had occurred
between the initial and after 120 min solutions.
The rate at which organic pollutants can be degraded by sono-
chemical (or other advanced oxidation) processes may be affected
by various parameters related to the generation of hydroxyl radi-
cals and other reactive oxygen species (ROS), the properties and
concentration of the pollutants and the water matrix itself.
For instance, both the ultrasound frequency and power density
will determine the number and size of collapsing cavitation bub-
bles, and eventually, the concentration of ROS in the liquid bulk
[23]. Of these, hydroxyl radicals are the dominant species formed
through water sonolysis:
H2OþÞÞÞ ! ÅH þ HOÅ
ðR1Þ
Moreover, the addition of SPS may promote degradation since
recent studies have shown that ultrasound can activate SPS to pro-
duce reactive sulfate radicals [15,24,25]:
S2O2ꢁþÞÞÞ ! 2SOꢁ4 Å
ðR2Þ
8
Finally, the reaction timescale was selected to achieve consider-
able EP degradation within a reasonable period of time, according
to screening experiments (i.e. Fig. 2).
Fig. 1. Response of the YES assay to EP (-O-) and E2 (-
D
-) at various concentrations.
Fig. 2. Sonochemical degradation profiles of EP at various operating conditions as
Inset: Triplicate wells of blanks, EP and E2.
shown in Table 2. -O- Run 7; -X- Run 30; -D- Run 20; -▲- Run 10.

C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
65
Table 1
Table 2
Independent parameters of the 26 experimental design.
Observed response Y (lg/L of EP removed) and pseudo-first order kinetic constant
(minꢁ1) for the 26 factorial experimental design. Numbers in brackets show linear
Value
level
EP
g/L)
Ultrasound
(US)
SPS
(mg/L)
pHo
Water
matrix
Time
(min)
regression coefficient R2 (%).
(
l
density
(W/L)
Run EP Ultrasound SPS pHo Water
no
Time
Y
kapp ꢀ 103
(US)
matrix
density
ꢁ1
250
1250
20
60
0
100
3
8
UPW
WW
30
120
+1
1a
1b
2a
2b
3a
3b
4a
4b
5a
5b
6a
6b
7a
7b
8a
8b
ꢁ
ꢁ
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
ꢁ
+
71.3 11 (98.9)
183.2
173.9 5.4 (98)
564
116.9 21.8 (98.6)
229.9
476.5 16.5 (98.9)
1066.8
85.7 12.6 (98.3)
203.4
174.1 6 (98.5)
570
135.4 25.3 (98.4)
239
521.6 19.1 (98.7)
1108.8
108.6 18.2 (98.8)
224.5
236.8 7.1 (98.7)
710.5
145.3 30.3 (98.8)
242.8
604.9 21.5 (97.5)
1160.8
116.9 19.8 (98.8)
229.9
295.8 9.4 (98.9)
825.5
159.8 35.6 (98.4)
245.8
ꢁ
It should be noted that the ultrasound frequency was not
selected as a process variable but kept fixed at 20 kHz due to lack
of equipment operating at higher frequencies. In recent studies,
various EDCs have been treated by both high (i.e. in the range
300–1200 kHz [14,18–20]) and low (i.e. 20–80 kHz [15–17]) fre-
quency ultrasound. High frequencies (at about 300 kHz) can pro-
duce more free radicals in the liquid bulk [23], thus oxidizing
faster hydrophilic and non-volatile compounds such as EP (whose
solubility is 885 mg/L and the vapor pressure is 0.01239 Pa at 25 °C
[1]). On the contrary, low frequencies yield more violent cavitation
with increased localized temperatures and pressures, thus promot-
ing thermal reactions inside and/or close to the bubble. Overall, the
effect of frequency is case-specific and interdependent to the
physicochemical properties of the substrate, the level of operating
power density and the relative concentration of ROS to substrate
(the latter will be discussed in Section 3.3).
+
+
ꢁ
ꢁ
+
ꢁ
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
+
+
+
ꢁ
+
+
9a
9b
ꢁ
ꢁ
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
+
ꢁ
+
10a
10b
11a
11b
12a
12b
13a
13b
14a
14b
15a
15b
16a
16b
17a
17b
18a
18b
19a
19b
20a
20b
21a
21b
22a
22b
23a
23b
24a
24b
25a
25b
26a
26b
27a
27b
28a
28b
29a
29b
30a
30b
31a
31b
32a
32b
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
Table 2 shows the 26 experimental design followed in this work,
alongside the response (dependent variable), Y, in terms of lg/L of
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
EP removed. It should be noted here that for experimental designs
where the pollutant concentration is one of the independent
parameters, it is not advisable to use the per cent conversion as
the response factor. Conversion, depending on the degradation
kinetics (see Section 3.3), will either remain constant or decrease
with increasing concentration although the mass of pollutant
destroyed may, at the same time, increase. From an environmental
perspective, the key factor should be the amount of pollutant
removed rather than its conversion.
Estimation of the average effect, the main effects and the two
and higher order interactions of the operating factors studied
was made by means of the statistical package Minitab 17. The
results are presented in Table 3. To assess the significance of the
effects, an estimate of the standard error is required. An estimate
of the standard error is usually made by performing repeat runs.
Alternatively, three and higher order interactions can be used,
since these interactions may be considered negligible and may
measure differences arising from experimental error [26]. The vari-
ance of each effect would then be:
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
+
+
+
ꢁ
687
26.8 (99.2)
+
+
1194.9
70.3 11.9 (97.9)
183.2
174.1 5.2 (98.2)
564
80.7 12.3 (98.3)
197.5
295.8 8.8 (98.7)
825.5
85.7 13.6 (98.5)
194.2
174.1 6 (97.8)
641.6
99.9 15.4 (98.6)
217.6
426.4 13.1 (99)
1017
108.6 20.4 (98.4)
224.4
351.4 10.6 (96.9)
916.1
80.7 25.3 (99.5)
297.6
543.1 18.5 (98.2)
1170.9
112.8 20.6 (98.7)
227.3
428.7 14.8 (98.9)
1017.8
142.1 30.5 (98.8)
241.7
ꢁ
ꢁ
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
+
ꢁ
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
+
+
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
+
+
+
+
+
ꢁ
+
P
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
2
ðthree and higher order effectÞ
ꢁ
Variance of effects ¼
Number of three and higher order effects
+
+
ꢁ
ꢁ
+
ꢁ
+
ðE1Þ
ꢁ
The standard error is then the square root of the variance (half
this amount for the average). If an effect is about or below the stan-
dard error, it may be considered insignificant. The contribution of a
parameter, however, whose effect appears different from zero, is
not necessarily very large. One way to identify the most important
effects is to construct the normal probability plot [26,27], where
small effects will appear on a straight line. Any effects with a sig-
nificant contribution will lie away from the normal probability
line. The normal probability plot for the degradation of EP appears
in Fig. 3a. With the exception of water matrix, the other five
parameters appear to have an effect on EP degradation; the effects
are positive indicating that an increase in their level brings about
an increase in the amount of EP removed. Second order interac-
tions between (EP concentration) ꢀ (US density), (EP concentra-
tion) ꢀ (pHo) and (EP concentration) ꢀ (time) also appear to
+
+
ꢁ
ꢁ
+
ꢁ
+
ꢁ
+
+
ꢁ
ꢁ
+
ꢁ
+
+
+
+
ꢁ
624.1 21.3 (98.6)
1173.5
+
+
influence degradation to a reasonable degree, while other interac-
tions lie closer to the normal probability line and their effect may
not be highly significant. This can evidently be illustrated in Fig. 3b
where the effect of all parameters and their interactions is shown
in the form of the Pareto chart; the ordinate shows the absolute

66
C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
Table 3
99
Estimated effects of the 26 factorial design for the removal of EP.
A
(a)
F
Effect
Value
95
90
AF
B
Average effect
406.5 1.85
AB
D
AD
Main effects
EP concentration
US density
SPS
pHo
Matrix
C
80
70
60
50
40
DE
ADE
AC
481.7 3.7
172.6 3.7
38.1 3.7
115.1 3.7
ꢁ6.3 3.7
318.8 3.7
BF
ABF
DF
Factor Name
A
B
C
D
E
Ethyl Paraben
Ultrasound
30
Time
Sodium persulfate
pH
20
Two-factor interactions
(EP concentration) ꢀ (US density)
(EP concentration) ꢀ (SPS)
(EP concentration) ꢀ X(pHo)
(EP concentration) ꢀ (Matrix)
(EP concentration) ꢀ (Time)
(US density) ꢀ (SPS)
(US density) ꢀ (pHo)
(US density) ꢀ (Matrix)
(US density) ꢀ (Time)
(SPS) ꢀ (pHo)
144.9 3.7
27.3 3.7
82.9 3.7
4.6 3.7
202.4 3.7
5.6 3.7
ꢁ11.4 3.7
ꢁ50.1 3.7
24.3 3.7
ꢁ0.8 3.7
8.2 3.7
ꢁ1.4 3.7
35.8 3.7
16.1 3.7
13.2 3.7
Water Matrix
Time
10
5
BDF
ABE
F
BE
1
-
10
10
0
20
30
40
50
60
70
Standardized Effect
Term
A
F
(b)
AF
(SPS) ꢀ (Matrix)
B
(SPS) ꢀ (Time)
AB
D
(pHo) ꢀ (Matrix)
AD
BE
(pHo) ꢀ (Time)
ABE
C
Reduced model
(Matrix) ꢀ (Time)
DE
ADE
AC
BF
Three-factor interactions
(EP concentration) ꢀ (US density) ꢀ (SPS)
(EP concentration) ꢀ (US density) ꢀ (pHo)
(EPconcentration) ꢀ (US density) ꢀ (Matrix)
(EP concentration) ꢀ (US density) ꢀ (Time)
(EP concentration) ꢀ (SPS) ꢀ (pHo)
(EP concentration) ꢀ (SPS) ꢀ (Time)
(EP concentration) ꢀ (pHo) ꢀ (Matrix)
(EP concentration) ꢀ (Matrix) ꢀ (Time)
(US density) ꢀ (SPS) ꢀ (pHo)
5.1 3.7
ꢁ9 3.7
ABF
BDF
DF
ꢁ41.3 3.7
21.8 3.7
4.5 3.7
BCD
EF
Factor Name
ADF
DEF
BCF
BD
CDE
CDF
ACE
ABD
CE
Full model
A
B
C
D
E
F
40
Ethyl Paraben
Ultrasound
7.4 3.7
29.7 3.7
4.2 3.7
Sodium persulfate
pH
Water Matrix
Time
50
ꢁ15.1 3.7
ACF
E
(US density) ꢀ (SPS) ꢀ (Matrix)
(US density) ꢀ (SPS) ꢀ (Time)
4.4 3.7
ꢁ11.5 3.7
ꢁ0.3 3.7
ꢁ16.6 3.7
ꢁ0.3 3.7
ꢁ11.1 3.7
ꢁ9.8 3.7
ꢁ1 3.7
0
20
30
60
70
10
(US density) ꢀ (pHo) ꢀ (Matrix)
(US density) ꢀ (pHo) ꢀ (Time)
(US density) ꢀ (Matrix) ꢀ (Time)
(SPS) ꢀ (pHo) ꢀ (Matrix)
Standardized Effect
Fig. 3. Normal probability plot (a) and Pareto chart (b) of the effects for EP removal.
White bars: effects of high significance; Black bars: effects of low significance; Gray
bars: Non-significant effects.
(SPS) ꢀ (pHo) ꢀ (Time)
(SPS) ꢀ (Matrix) ꢀ (Time)
(pHo) ꢀ (Matrix) ꢀ (Time)
11.7 3.7
The coefficients that appear in Eq. (E2) are half the calculated
effects of Table 3, since a change from the low value (ꢁ1) to the
upper value (+1) is a change of two units along the axis. The model
predicts a linear dependency of the mass of EP removed on the
operating parameters and the respective interactions. Usually,
the three and higher order interactions are not expected to be sig-
nificant. Their existence in the model indicates that the response
surface is actually non-linear.
One of the objectives of the experimental design is to provide a
simple and reliable model capable of relating directly the response
factor to the most significant parameters. As seen in Fig. 3b, of the
various effects that are statistically significant and appear in Eq.
(E2), those shown as black bars are definitely less important than
the rest (shown as white bars). To assess whether the model repre-
sented by Eq. (E2) may be simplified omitting these ten terms, the
values of the residuals (i.e. observed minus predicted values of
response) were plotted in a normal probability plot for both the full
and the reduced models (Fig. 4). As seen, most of the points from
the residual plot for the reduced model lie close to the straight line
(confidence interval = 95%) confirming the conjecture that effects
associated with the least significant terms of Eq. (E2) may be
explained by random noise. The regression coefficient (R2) is
98.4% and 97.2% for the full and reduced model, respectively.
Therefore, the experimental response of EP removal can be reduced
and described as follows:
value of the effect, while the abscissa shows the main effects and
the interactions. The vertical line shown in Fig. 3b is a reference
line indicating the Lenth’s pseudo-standard error; any effect that
extends past this line is potentially significant [28].
Based on the parameters and interactions which are statistically
significant (i.e. represented by white and black bars in Fig. 3b), a
full model describing the experimental response was constructed
as follows:
Y ¼ 406:5 þ 240:9 EP concentration þ 86:3 US density
þ 19:1 SPS þ 57:6 pHo þ 159:4 Time
þ 72:5 EP concentration ꢂ US density
þ 13:7 EP concentration ꢂ SPS þ 41:5 EP concentration
ꢂ pHo þ 101:2 EP concentration ꢂ Time
ꢁ 25:1 US density ꢂ Matrix þ 12:2 US density ꢂ Time
þ 17:9 pHo ꢂ Matrix þ 8:1 pHo ꢂ Time
ꢁ 20:7 EP concentration ꢂ US density ꢂ Matrix
þ 10:9 EP concentration ꢂ US density ꢂ Time
þ 14:9 EP concentration ꢂ pHo ꢂ Matrix
ꢁ 8:3 US density ꢂ pHo ꢂ Time
ðE2Þ

C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
67
Y ¼ 406:5 þ 240:9 EP concentration þ 86:3 US density
þ 19:1 SPS þ 57:6 pHo þ 159:4 Time
reported [16] that estrogen sonodegradation in WW was faster
than in UPW at pH 8 but the opposite occurred at pH 3, which
shows that interactions might be significant. Sasi et al. [20]
reported that methyl paraben sonochemical degradation was
severely retarded in the presence of carbonates, while other anions
such as chlorides, nitrates and sulfates had no effect. They also
reported decreased degradation rates when methyl paraben was
co-treated with other EDCs such as propyl paraben, phthalates
and triclosan.
Regarding the effect of pH on EP sonochemical degradation, the
rate was found to increase from acidic pH 3 to slightly alkaline pH
8. In general, the pH of the reaction medium has a complex effect
on the sonochemical degradation rates of organic pollutants
depending on the net charge (state) of the pollutant and the avail-
able amount of hydroxyl radicals and hydrogen peroxide (formed
through radicals recombination) to attack the target pollutant
[31,32]. Although at low pH values, EP molecules exist in non-
ionic form (pKa = 8.22 [1]) and can easily approach the negatively
charged cavity bubbles, better degradation efficiency of EP was
observed in alkaline conditions. This can be attributed to the for-
mation of stable H3O+2 cations resulting from the protonation of
hydrogen peroxide, as well as to the scavenging of HOÅ radicals
by H+ at low pH values, limiting the degradation rate [31,32]. Thus,
pH 8 could be considered as the reaction condition that counter-
balances better both the formation of HOÅ radicals and the molec-
ular form of EP, which can diffuse to the negatively charged
hydrophobic interface of liquid–gas cavitation bubbles compared
with pH 3.
þ 72:5 EP concentration ꢂ US density
þ 13:7 EP concentration ꢂ SPS þ 41:5 EP concentration
ꢂ pHo þ 101:2 EP concentration ꢂ Time
ðE3Þ
The reduced model consists of seven positive effects, of which
EP concentration is the most important one followed by the reac-
tion time and their interaction, (EP concentration) ꢀ (time). The
beneficial effect of increasing power density has to do with an
increase in the number of collapsing bubbles, thus leading to
enhanced degradation levels [23]. The relatively low but still statis-
tically significant effect of SPS (which appears in the full but not in
the reduced model) may be associated with the fact that ultra-
sound can only partially activate SPS, as will be seen in Section 3.4.
Interestingly and rather unexpectedly, the individual (main) effect
of water matrix is insignificant at the conditions in question
although higher order interactions such as (US density) ꢀ (matrix),
(pHo) ꢀ (matrix), (US density) ꢀ (matrix) ꢀ (EP concentration) and
(pHo) ꢀ (matrix) ꢀ (EP concentration) appear in the full model. The
effluent organic matter, as well as the various anions typically
found in WW are known to influence drastically the degradation
kinetics of micro-pollutants. The impact is typically adverse
[15,29,30] since the WW components behave as ROS scavengers
competing with the substrate. Nonetheless, we have previously
99.9
(a)
3.3. Kinetics
99
Although a detailed kinetic analysis was outside the scope of
this work, an attempt was made to evaluate the reaction order
with respect to EP. Besides the sampling times of 30 and 120 min
used for factorial analysis, six extra samples were also taken at
other intervals to obtain the EP concentration–time profiles for
all the experiments (i.e. like those shown in Fig. 2). The respective
data were then fitted to a pseudo-first order kinetic expression, as
follows:
95
90
80
70
60
50
40
30
20
10
5
ꢁdC=dt ¼ kappC () LnðC=CoÞ ¼ ꢁkappt () Lnð1 ꢁ XÞ
¼ ꢁkappt
ðE4Þ
1
where kapp is an apparent rate constant and X is the conversion. The
last column of Table 2 shows the computed rate constants, along-
side the regression coefficients of linear fitting. Comparing experi-
ments 1 and 2, 3 and 4, 5 and 6 and so on (i.e. experiments
performed at low and high EP concentrations, while all other
parameters are the same), kapp values change with EP concentration,
which implies that the rate is not true first order regarding EP con-
centration. The fact that kapp consistently decreases with increasing
concentration reveals that the reaction order is actually between
one and zero. This can better be seen comparing conversion values.
0.1
-200
-100
0
100
200
Residuals - Full model
99.9
99
(b)
95
90
80
70
60
50
40
30
For the run 1 performed at 250
conversion is 28.5% and 73.3%, respectively, while for the run 2 per-
formed at 1250 g/L EP, the respective values are 13.9% and 45.2%.
lg/L EP, the 30-min and 120-min
l
If the reaction were true zero order, i.e.:
20
10
5
ꢁdC=dt ¼ kapp () Co ꢁ C ¼ kappt () X ¼ kappt=Co
ðE5Þ
a 5-fold increase in concentration would result in a 5-fold decrease
in conversion. The fact that conversion and the rate constant
decrease as concentration increases denotes kinetics between zer-
oth and first order.
At fixed operating conditions (i.e. ultrasound frequency and
power density, liquid bulk temperature), the rate of ROS generation
will be nearly constant and the kinetics of EP degradation will be
1
0.1
-200
-100
0
100
200
Residuals - Reduced model
Fig. 4. Normal probability plots of the residuals for EP removal. Full model (a) and
reduced model (b).

68
C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
Table 4
1.4E+05
1.2E+05
1.0E+05
8.0E+04
6.0E+04
4.0E+04
2.0E+04
0.0E+00
35
High resolution accurate mass data ([M-H]ꢁ, MS product ions and relative error
(ppm) for EP and TBPs.
D
3,4-DHB
4-HB
30
25
20
15
10
5
2,4 –DHB
MP
Rt (min) TBP
Pseudo-
m/z
D (ppm)
molecular [M-H]ꢁ
formula
7.9
6.9
1.4
Ethyl paraben (EP)
C9H9O3
C8H7O3
C7H5O4
165.0551 3.4
151.0401 1.1
153.0200 4.5
Methyl paraben (MP)
3,4 dihydroxybenzoic acid
(3,4-DHB)
1.6
2.5
2,4 dihydroxybenzoic acid
(2,4-DHB)
C7H5O4
153.0200 4.5
4 hydroxybenzoic acid (4-HB) C7H5O3
137.0244 ꢁ2.4
0
0
1
2
3
4
5
6
Time, h
dictated by the ROS to EP concentration ratio. If this ratio is low
(which may occur at high EP concentrations and/or conditions that
do not favor the rapid ROS production), the reaction will approach
zeroth order kinetics; conversely, a shift toward higher order
kinetics is expected at higher ROS to EP ratios.
Fig. 5. Evolution profiles of TBPs during the sonochemical degradation of 3.5 mg/L
EP in UPW, 20 W/L and 500 mg/L SPS. MP concentration is shown in secondary axis
in terms of chromatographic peak area.
3.4. Identification of TBPs and degradation pathways
To identify TBPs, an additional 8-h long experiment was per-
formed in the presence of 500 mg/L SPS and at an increased EP con-
centration of 3.5 mg/L to facilitate chromatographic analysis. The
formation of TBPs was evaluated by means of LC–TOF–MS. Analy-
ses were performed on pre-concentrated samples in full scan mode
in order to facilitate the detection of as many TBPs as possible.
Total ion chromatograms and filtered m/z chromatograms
obtained after full-scan analyses of pre-concentrated samples indi-
cated the formation of four TBPs (Table 4). Identification of TBPs
(with the exception of methyl paraben) was also verified by com-
parison of retention times and MS spectra of the authentic refer-
ence standards in a second step.
One transformation product of EP with a molecular formula of
C8H7Oꢁ3 and pseudo-molecular [M-H]ꢁ ion at m/z 151.0401 was
attributed to a dealkylated TBP and more specifically to methyl
paraben (MP). Further dealkylation leads to the formation of one
TBP eluting at 6.9 min with [M-H]ꢁ m/z 137.0244 and elemental
formula of C7H5Oꢁ3 . Based on both literature data and comparison
by authentic standards, TBP with m/z 137.0244 is attributed to
4-hydroxybenzoic acid (4-HB).
Two more TBPs with isomeric structure were identified as dihy-
droxybenzoic acid (DHB) derivatives namely 2,4-DHB (m/z
153.0200) and 3,4-DHB (m/z 153.0200). Their identity was con-
firmed with the analysis of commercially available standards. Sim-
ilar TBPs have been previously identified during the photocatalytic
treatment of EP [12]. In agreement with our results, dealkylated
and dihydroxybenzoic acids were the major TBPs during the treat-
ment of compounds belonging also to the chemical category of
para-hydroxybenzoates such as methyl and benzyl paraben by dif-
ferent advanced oxidation AOPs [33–35].
Fig. 6. Evolution of sulfate ion (-d-), EP (-O-) and toxicity (-▲-) at the conditions of
Fig. 5.
SOÅ4ꢁ attack followed by addition of dissolved O2 and decarboxyla-
tion leads to dealkylated MP [15,35–40].
TBPs evolution as a function of treatment time was also fol-
lowed. Fig. 5 shows the concentration–time profiles of 4-HB, 2,4-
DHB and 3,4-DHB, whereas for MP the chromatocgraphic peak area
as a function of reaction time is depicted. All of them were readily
transformed and completely removed after 6 h of ultrasound irra-
diation. Based on the profiles of Fig. 5, MP was found to be a first
stage TBP of EP degradation attaining its maximum concentration
within 1 h. 4-HB attained its maximum concentration within 2 h
proving the consecutive steps of dealkylation. On the other hand,
the formation of dihydroxybenzoic acids peak at longer irradiation
times (4 h).
Fig. 6 shows the evolution of EP, sulfate ion and toxicity during
the reaction; toxicity to V. fischeri decreases from 62% to 25% after
4 h of reaction and this coincides with 90% EP removal, thus imply-
ing that TBPs are less toxic than the parent compound. Interest-
ingly, the measured concentration of sulfate ion at the end of the
experiment corresponds to only 21% utilization of SPS and its
sono-activated conversion to sulfate radicals.
It is documented in the literature that HOÅ and SO4Åꢁ can lead to
the formation of similar TBPs though different mechanisms
[15,36,37]. More specifically, hydroxylated-TBPs can be formed
via the following mechanistic routes: (i) electrophilic attack of
HOÅ radicals, and (ii) electron transfer from the aromatic ring to
the SOÅ4ꢁ generating radical cations. Subsequent reaction of radical
cations with H2O can generate hydroxycyclohexadienyl radicals
through addition or elimination pathways. In both cases, the pres-
ence of electron-donating HO group in the substituted benzenes
favors subsequent hydroxylation [15,36–39]. Similarly, dealkylated
TBPs (i.e. MP) can be formed through the direct loss of –CH3 group
of ethyl chain by HOÅ and SO4Åꢁ attack. Alternatively, radical forma-
tion by hydrogen abstraction from the –CH3 group after HOÅ and
Based on the detected TBPs and their evolution profiles, the son-
odegradation pathway of EP is proposed in Fig. 7, including succes-
sive dealkylation of ethyl chain, followed by the hydroxylation of
the aromatic ring leading to the formation of dihydroxybenzoic
acids. This implies that hydroxyl radical-driven reactions play an
important role in the process; these reactions are likely to occur
in the neighborhood of the cavitation bubble and/or in the liquid

C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
69
O
C
O
C
O
O
H
O SO
,
4
HO
HO
MP
EP
O SO
,
H
4
O
C
O
C
OH
O
C
HO
HO
H
O
OH
OH
OH
+
HO
HO
-
4 HB
-
3 4
,
DHB
-
2 4 DHB
,
Fig. 7. Sonochemical degradation pathways of EP.
bubble, while thermal degradation inside the bubble is unlikely
to happen given that EP is a water soluble and non-volatile species.
All the above intermediates can be further transformed by an
oxidative opening of the aromatic ring via HOÅ and/or SO4Åꢁ contin-
uous attack, giving rise to the formation of lower molecular weight
and more oxidized molecules e.g. oxalic, formic, acetic, lactic,
malonic, tartaric, glyceric and succinic acid, as reported elsewhere
[36].
Acknowledgement
This work was supported by Grant E056 from the Research
Committee of the University of Patras (Program C. Caratheodory).
References
[1] C. Haman, X. Dauchy, C. Rosin, J.F. Munoz, Occurrence, fate and behavior of
parabens in aquatic environments: a review, Water Res. 68 (2015) 1–11.
[2] E.J. Routledge, J. Parker, J. Odum, J. Ashby, J.P. Sumpter, Some alkyl hydroxy
benzoate preservatives (parabens) are estrogenic, Toxicol. Appl. Pharmacol.
153 (1998) 12–19.
4. Conclusions
Parabens constitute a new family of emerging micro-pollutants,
whose advanced oxidation has merely been investigated so far.
This is particularly true for their sonochemical degradation and,
in this perspective, ethyl paraben was chosen as a representative
compound to be studied. The main conclusions of this work are
as follows:
[3] S. Oishi, Effects of butylparaben on the male reproductive system in rats,
Toxicol. Ind. Health 217 (2001) 31–39.
[4] P.D. Darbre, A. Aljarrah, W.R. Miller, N.G. Coldham, M.J. Sauer, G.S. Pope,
Concentrations of parabens in human breast tumours, J. Appl. Toxicol. 24
(2004) 5–13.
[5] D. Bledzka, J. Gromadzinska, W. Wasowicz, Parabens. From environmental
studies to human health, Environ. Int. 67 (2014) 27–42.
[6] I. Gonzalez-Marino, J.B. Quintana, I. Rodriguez, R. Cela, Evaluation of the
occurrence and biodegradation of parabens and halogenated by-products
in wastewater by accurate-mass liquid chromatography-quadrupole-time-
of-flight-mass spectrometry (LC–QTOF–MS), Water Res. 45 (2011) 6770–
6780.
(1) To evaluate the importance of operating parameters on
degradation, a factorial design approach was implemented.
Six parameters, namely ethyl paraben concentration, reac-
tion time, ultrasound power, initial pH, water matrix and
the addition of sodium persulfate (as a source of sulfate rad-
icals) were tested. Of primary importance were the first four,
as well the second order interactions between concentration
and each of the other three parameters. The effect of persul-
fate, although statistically significant, was less important
than the others, while the water matrix effect was non-
significant.
[7] J.R. Steter, R.S. Rocha, D. Dionisio, M.R.V. Lanza, A.J. Motheo, Electrochemical
oxidation route of methyl paraben on
Electrochim. Acta 117 (2014) 127–133.
a boron-doped diamond anode,
[8] L. Hernandez-Leal, H. Temmink, G. Zeeman, C.J.N. Buisman, Removal of
micropollutants from aerobically treated grey water via ozone and activated
carbon, Water Res. 45 (2011) 2887–2896.
[9] D. Dobrin, M. Magureanu, C. Bradu, N.B. Mandache, P. Ionita, V.I. Parvulescu,
Degradation of methylparaben in water by corona plasma coupled with
ozonation, Environ. Sci. Pollut. Res. 21 (2014) 12190–12197.
[10] D. Bledzka, D. Gryglik, M. Olak, J.L. Gebicki, J.S. Miller, Degradation of n-
butylparaben and 4-tert-octylphenol in H2O2/UV system, Radiat. Phys. Chem.
79 (2010) 409–416.
(2) An empirical mathematical model capable of simulating
quantitatively the amount of paraben removed was devel-
oped. The model contains the aforementioned four individ-
ual effects and the three second order interactions, while
effects of lower significance were omitted to obtain a simpli-
fied (reduced) model without losing accuracy.
(3) Such models should be viewed cautiously since they are
applicable within the range of conditions they have been
developed for. For this range, the rate of ethyl paraben sono-
chemical degradation obeys kinetics between 0th and 1st
order with respect to concentration.
[11] P. Atheba, P. Drogui, B. Seyhi, D. Robert, Photo-degradation of butyl
parahydroxybenzoate by using TiO2-supported catalyst, Water Sci. Technol.
67 (2013) 2141–2147.
[12] A. Petala, Z. Frontistis, M. Antonopoulou, I. Konstantinou, D.I. Kondarides, D.
Mantzavinos, Kinetics of ethyl paraben degradation by simulated solar
radiation in the presence of N-doped TiO2 catalysts, Water Res. 81 (2015)
157–166.
[13] T. Velegraki, E. Hapeshi, D. Fatta-Kassinos, I. Poulios, Solar-induced
heterogeneous photocatalytic degradation of methyl-paraben, Appl. Catal. B
Environ. 178 (2015) 2–11.
[14] R.A. Torres, C. Petrier, E. Combet, M. Carrier, C. Pulgarin, Ultrasonic cavitation
applied to the treatment of bisphenol A. Effect of sonochemical parameters
and analysis of BPA by-products, Ultrason. Sonochem. 15 (2008) 605–611.
[15] B. Darsinou, Z. Frontistis, M. Antonopoulou, I. Konstantinou, D. Mantzavinos,
Sono-activated persulfate oxidation of bisphenol A: Kinetics, pathways and the
controversial role of temperature, Chem. Eng. J. 280 (2015) 623–633.
(4) During the early steps of the reaction, ethyl paraben under-
goes dealkylation to form methyl paraben, whose subse-
quent hydroxylation yields mono and dihydroxybenzoic
acids.
[16] Z. Frontistis, D. Mantzavinos, Sonodegradation of 17a-ethynylestradiol in
environmentally relevant matrices: laboratory-scale kinetic studies, Ultrason.
Sonochem. 19 (2012) 77–84.

70
C. Papadopoulos et al. / Ultrasonics Sonochemistry 31 (2016) 62–70
[17] G. Andaluri, E.V. Rokhina, R.P.S. Suri, Evaluation of relative importance of
ultrasound reactor parameters for the removal of estrogen hormones in water,
Ultrason. Sonochem. 19 (2012) 953–958.
[18] M. Chiha, O. Hamdaoui, S. Baup, N. Gondrexon, Sonolytic degradation of
endocrine disrupting chemical 4-cumylphenol in water, Ultrason. Sonochem.
18 (2011) 943–950.
[30] L. Sanchez-Prado, R. Barro, C. Garcia-Jares, M. Llompart, M. Lores, C. Petrakis, N.
Kalogerakis, D. Mantzavinos, E. Psillakis, Sonochemical degradation of
triclosan in water and wastewater, Ultrason. Sonochem. 15 (2008) 689–694.
[31] M.V. Bagal, P.R. Gogate, Wastewater treatment using hybrid treatment
schemes based on cavitation and Fenton chemistry:
Sonochem. 21 (2014) 1–14.
a review, Ultrason.
[19] L.J. Xu, W. Chu, N. Graham, A systematic study of the degradation of dimethyl
phthalate using a high-frequency ultrasonic process, Ultrason. Sonochem. 20
(2013) 892–899.
[32] M. Lastre-Acosta, G. Cruz-González, L. Nuevas-Paz, U.J. Jáuregui-Haza, A.C.
Silva Costa Teixeira, Ultrasonic degradation of sulfadiazine in aqueous
solutions, Environ. Sci. Pollut. Res. 22 (2015) 918–925.
[20] S. Sasi, M.P. Rayaroth, D. Devadasan, U.K. Aravind, C.T. Aravindakumar,
Influence of inorganic ions and selected emerging contaminants on the
degradation of methylparaben: a sonochemical approach, J. Hazard. Mater.
300 (2015) 202–209.
[21] R. Daghrir, A. Dimboukou-Mpira, B. Seyhi, P. Drogui, Photosonochemical
degradation of butyl-paraben: optimization, toxicity and kinetic studies, Sci.
Total Environ. 490 (2014) 223–234.
[33] X. Feng, Y. Chen, Y. Fang, X. Wang, Z. Wang, T. Tao, Y. Zuo, Photodegradation of
parabens by Fe(III)-citrate complexes at circumneutral pH: matrix effect and
reaction mechanism, Sci. Total Environ. 472 (2014) 130–136.
[34] Y. Gao, T. An, H. Fang, Y. Ji, G. Li, Computational consideration on advanced
oxidation degradation of phenolic preservative, methylparaben, in water:
mechanisms, kinetics, and toxicity assessments, J. Hazard. Mater. 278 (2014)
417–425.
[22] E.J. Routledge, J.P. Sumpter, Estrogenic activity of surfactants and some of their
degradation products assessed using a recombinant yeast screen, Environ.
Toxicol. Chem. 15 (1996) 241–248.
[35] Y. Lin, C. Ferronato, N. Deng, J.M. Chovelon, Study of benzylparaben
photocatalyticdegradation by TiO2, Appl. Catal. B Environ. 104 (2011) 353–
360.
[23] Y.G. Adewuyi, Sonochemistry: environmental science and engineering
applications, Ind. Eng. Chem. Res. 40 (2001) 4681–4715.
[24] S. Wang, N. Zhou, S. Wu, Q. Zhang, Z. Yang, Modeling the oxidation kinetics of
sono-activated persulfate’s process on the degradation of humic acid, Ultrason.
Sonochem. 23 (2015) 128–134.
[25] J.C. Lin, S.L. Lo, C.Y. Hu, Y.C. Lee, J. Kuo, Enhanced sonochemical degradation of
perfluorooctanoic acid by sulfate ions, Ultrason. Sonochem. 22 (2015) 542–
547.
[26] G.E.P. Box, W.G. Hunter, J.S. Hunter, Statistics for Experimenters, John Wiley &
Sons, New York, 1978.
[27] C. Daniel, Applications of Statistics to Industrial Experiments, John Wiley &
Sons, New York, 1976.
[36] R. Hazime, Q.H. Nguyen, C. Ferronato, A. Salvador, F. Jaber, J.-M. Chovelon,
Comparative study of imazalil degradation in three systems: UV/TiO2, UV/
K2S2O8 and UV/TiO2/K2S2O8, Appl. Catal. B Environ. 144 (2014) 286–291.
[37] R. Hazime, Q.H. Nguyen, C. Ferronato, F. Jaber, J.-M. Chovelon, Optimization of
imazalil removal in the system UV/TiO2/K2S2O8 using a response surface
methodology (RSM), Appl. Catal. B Environ. 132–133 (2013) 519–526.
[38] M. Antonopoulou, I. Konstantinou, TiO2 photocatalysis of 2-isopropyl-3-
methoxy pyrazine taste and odor compounds in aquatic systems: kinetics,
degradation pathways and toxicity evaluation, Catal. Today 240 (2015) 22–29.
[39] M.G. Antoniou, A. De La Cruz, D.D. Dionysiou, Intermediates and reaction
pathways from the degradation of microcystin-LR with sulfate radicals,
Environ. Sci. Technol. 44 (2010) 7238–7244.
[28] R.V. Lenth, Quick and easy analysis of unreplicated factorials, Technometrics
31 (1989) 469–473.
[29] B. Li, L. Li, K. Lin, W. Zhang, S. Lu, Q. Luo, Removal of 1,1,1-trichloroethane from
aqueous solution by a sono-activated persulfate process, Ultrason. Sonochem.
20 (2013) 855–863.
[40] M.V.P. Sharma, V. Durgakumari, M. Subrahmanyam, Solar photocatalytic
degradation of isoproturon over TiO2/H-MOR composite systems, J. Hazard.
Mater. 160 (2008) 568–575.