You are on page 1of 11

Journal of Environmental Chemical Engineering 9 (2021) 106442

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering


journal homepage: www.elsevier.com/locate/jece

Degradation of hormones in tap water by heterogeneous solar


TiO2-photocatalysis: Optimization, degradation products identification,
and estrogenic activity removal
Rodrigo Nogueira Padovan a, Lucas Sponton de Carvalho b, Patrícia Luisa de Souza Bergo d,
Chubraider Xavier a, *, Andrei Leitão c, Álvaro José dos Santos Neto b, Fernando Mauro Lanças b,
Eduardo Bessa Azevedo a
a
University of São Paulo (USP), Chemistry Institute of São Carlos (IQSC), Laboratório de Desenvolvimento de Tecnologias Ambientais (LDTAmb), Avenida Trabalhador
São-Carlense, 400 – Centro, P. O. Box: 780, São Carlos 13560-970, SP, Brazil
b
University of São Paulo (USP), Chemistry Institute of São Carlos (IQSC), Grupo de Cromatografia (CROMA), Avenida Trabalhador São-Carlense, 400 – Centro, P. O.
Box: 780, São Carlos 13560-970, SP, Brazil
c
University of São Paulo (USP), São Carlos Institute of Chemistry (IQSC), Medicinal & Biological Chemistry Group (NEQUIMED), Avenida Trabalhador São-Carlense,
400 – Centro, P. O. Box: 780, São Carlos 13560-970, SP, Brazil
d
Federal University of São Carlos (UFSCar), Department of Chemistry (DQ), Rodovia Washington Luiz, km 235, P.O. Box: 676, São Carlos 13565-905, SP, Brazil

A R T I C L E I N F O A B S T R A C T

Editor: Dr. G.L. Dotto Hormones are one of the most hazardous classes of Contaminants of Emerging Concern (CEC) found in tap water,
being significantly recalcitrant regarding traditional water treatment processes. In this work, the hormones
Keywords: estrone, 17β-estradiol, 17α-ethinylestradiol, and estriol, present in tap water, were degraded by heterogeneous
estrone solar photocatalysis in a flat-plate photochemical reactor (recycle mode) using TiO2 (P25, Evonik) as the pho­
17β-estradiol
tocatalyst. A fully automated analytical method based on online SPE and LC-UV was developed and validated for
17α-ethinylestradiol
monitoring those molecules. For a fixed amount of sample (125 μL), the limit of quantification was 10 μg L− 1 (for
estriol
estrogenic activity all four hormones) with coefficients of variation smaller than 20%, and accuracy between 80% and 120%, all of
MCF-7 which are acceptable figures for this kind of determination. The reactional system was optimized with the aid of a
22 factorial design. At optimized conditions (pH ≈ 6.8, flow rate 250 mL min− 1), the removal of the four hor­
mones (250 μg L− 1 each) was approximately 85% after 3 h of treatment. However, even after 9 h of treatment, it
was not possible to completely remove the estrogenic activity (MCF-7 cells) of the treated tap water. The
structures of some degradation products were proposed.

1. Introduction essentially, on the physical-chemical process of floc-coagulation and


decantation, followed by disinfection. However, those processes are not
Water treatment technology is a critical topic of discussion, as the use able to completely remove the so-called Contaminants of Emerging
of water resources defines how the whole society and the environment Concern (CECs), once they are not meant for that. CECs are species that
develop. The depletion of freshwater is severe in Africa and Asia, and the occur in the environment at concentrations ranging from ng L− 1 to μg
United Nations Educational, Scientific and Cultural Organization L− 1 and are not currently monitored by environmental agencies,
(UNESCO) estimates that only 39% of the world population has access to although their chronic effects are being reported in the literature [2].
safely managed drinking water services, while 48% has no access to Several CECs (for instance, drugs and natural hormones excreted by
water with any kind of treatment at all [1]. The traditional treatment of the population) act as endocrine disruptors (ED). When unwillingly
urban waters is focused on the removal of suspended solids and is based, ingested, those chemicals may be responsible for anomalies in the

* Corresponding author.
E-mail addresses: rodrigopadovan@iqsc.usp.br (R.N. Padovan), lucassponton@gmail.com (L.S. de Carvalho), patriciabergo@gmail.com (P.L. de Souza Bergo),
chubraiderxavier@usp.br (C. Xavier), andleitao@iqsc.usp.br (A. Leitão), alvarojsn@iqsc.usp.br (Á.J. dos Santos Neto), flancas@iqsc.usp.br (F.M. Lanças), bessa@
iqsc.usp.br (E.B. Azevedo).

https://doi.org/10.1016/j.jece.2021.106442
Received 4 August 2021; Received in revised form 16 September 2021; Accepted 21 September 2021
Available online 24 September 2021
2213-3437/© 2021 Elsevier Ltd. All rights reserved.
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 1. Scheme of the used SPE-online chromatograph.

photocatalyst that presents low cost, low ecotoxicity, and large pH range
usability [8]. TiO2 exhibits three possible crystalline phases: brookite,
the hardest one to be obtained in a pure state; rutile, the one with the
lowest photocatalytic activity; and anatase, the most photoactive one [9,
10]. Despite the higher activity of anatase, literature reports that Evo­
nik’s TiO2 (P25), a mixture of 70:30 anatase:rutile, gives superior re­
sults. Some authors argue that this effect is partially due to the formation
of heterojunctions between those two phases [11].
There are two main ways of working with TiO2 particles in photo­
catalytic experiments: in suspension or immobilized on a solid substrate.
The former leads to a higher specific surface area, but requires a further
step to recover the photocatalyst; the latter, although no separation is
needed, presents a considerably lower specific surface area [12].
Another approach aiming at decreasing treatment costs is the use of
solar-based AOPs. Although TiO2 is only excited by photons whose
Fig. 2. Scheme of the flat-plate photochemical reactor used in the solar pho­ wavelengths are greater than 388 nm (UV-A), corresponding to
tocatalytic experiments. Note: a) TiO2-coated glass plate over which water approximately 4% of the incident solar radiation at the Earth surface,
flows; b) hormone-contaminated water reservoir; c) magnetic stirrer; d) recycle several works show the feasibility of this kind of system for the removal
pump; e) bypass; f) flow control (valve and rotameter); g) reservoir filled with
of many compounds, hormones included [13–15].
replacement distilled water; h) solenoid valve for distilled water replacement; i)
Although degradation studies of CECs, like hormones, in real
automation device for distilled water replacement; j) contaminated-water
matrices, are of utmost importance, they are difficult to be performed
level sensor.
due to the low concentrations involved and severe interferences from
the matrix. This is possible nowadays thanks to advances in analytical
Table 1 instrumentation for chromatography and mass spectrometry [16,17].
Factors studied with coded and real values in the 22 full factorial design. The full automatization of the analytical process, with online extraction,
Factors Levels
makes it possible to work with a significant throughput. Another useful
approach for complex analyses is column switching: it is based on the
(–1) (+1)
use of valves, changing the path of the analytes through the chromato­
x1 : pH 4.5 8.0 graph. It makes it possible to use an extractor column or multidimen­
x2 : Flow rate (mL min− 1) 150 300 sional separation processes [18].
Therefore, one of the objectives of this work was to develop a fully
automated analytical method and use it in the optimized degradation of
metabolism [3]. ED-laden water consumption has been reported, in the estrone, 17β-estradiol, 17α-ethinylestradiol, and estriol, in tap water.
literature, as one of the causes for several types of cancer and develop­ Those hormones were degraded by heterogeneous solar photocatalysis
mental delay in children [4,5]. Estrone (E1), 17β-estradiol (E2), estriol in a TiO2-coated flat-plate photochemical reactor, following a 22 facto­
(E3), and 17α-ethinylestradiol (EE2) are of worldwide concern [6], as rial design. The degradation/energy ratio profile was obtained, some
they may have deleterious effects even at low concentrations. For degradation products were identified by mass spectrometry, and the
instance, few days of exposure to aqueous E2 (1 ng L− 1) or EE2 (0.1 ng estrogenic activity after treatment was determined using the MCF-7 cell
L− 1) caused fish to change their sexes [7]. line. Those are hormone-dependent human-breast cancer cells, in which
In this context, alternative water treatment technologies have been cell growth is driven by the addition of estradiol, previously described in
developed as an add-on to the traditional ones. That is the case of the [19].
advanced oxidation processes (AOPs), which are based on the in-situ
generation of radical species, aiming at oxidizing the target-molecules.
Among them, heterogeneous photocatalysis has risen as a prominent
technique, especially with titanium dioxide (TiO2): a recyclable

2
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Table 2
The four analyzed hormones (SIGMA-ALDRICH, 95% of purity).
Name Structure Molar mass (g mol− 1) Water solubility (mg L− 1
at 20ºC) logDpH=7.4 pKa

Estrone (E1) 270.37 5.03 4.31 10.771

17β-Estradiol (E2) 272.38 13.03 3.753 10.711

17α-Ethinylestradiol (EE2) 296.40 4.83 4.153 10.333

Estriol (E3) 288.38 13.03 2.673 10.42

Note: 1Lewis; Archer [25]; 2Marques et al. [26]; 3Chemaxon [27].

2. Materials and methods Supplementary Material.

2.1. Online-SPE columns manufacture


2.3. Identification of degradation products

The extraction columns were manufactured using stainless steel


Degradation products were determined by LC-MS/MS experiments.
tubes (60 mm length, 0.500 mm internal diameter). They were filled
An HPLC Shimadzu Serie 20 A Prominence™ (Shimadzu, Japan) with a
with 10 mg StrataX™ phase (Phenomenex, USA) suspended in 95:5
Kinetex™ XB-C18 (100 × 2.1 mm, 2.6 µm Phenomenex, USA) was used.
water:methanol (in volume) solution. The suspension was pumped
The separations were performed in reverse mode, gradient elution, with
(methanol 0.5 mL min− 1 for 15 min) into the stainless tube, which was
water (A) and acetonitrile (B): 0–3 min, 5% B; 3–12 min, 5–60% B;
closed by a 10.0 µm pore size stainless steel screen filter (VICI VALCO,
12–14 min, 60–95% B; 14–20 min, 95% B; 20–22 min, 95–5% B;
Switzerland) grid at the exit end. Finally, the particles (33 µm average
22–26 min, 5% B (0.25 mL min− 1 flow rate, 40ºC column heater
diameter and 800 m2 g− 1 specific surface area) were contained by
temperature).
another screen filter (at the other end) and a pair of end fittings. The SPE
A micrOTOF-Q II (Bruker Daltonics, USA) was coupled with the
columns were conditioned with 60 μL min− 1 of water for 60 min before
HPLC. The analyses conditions were: electrospray ionization source
use.
(capillary voltage 4.5 kV, desorption temperature 200ºC for both nega­
tive and positive modes), dryer gas flow rate 8 L min− 1, nebulizer
pressure 4 bar, m/z range 50–3000 u, acquisition rate 2 Hz, full MS
2.2. Liquid chromatography analyses
mode.
The chromatographic setup used is shown in Fig. 1. An UltiMate
3000 RSLCnano (Thermo Scientific, USA) chromatograph was used for 2.4. Photochemical reactor
this assembling.
The chromatograph was composed of: 125-μL sample loop, six-port A flat-plate photochemical reactor (Fig. 2) was manufactured ac­
valve, ternary pump for carrying the sample using water as the solvent cording to Vilela et al. [20] using TiO2 (P25, Evonik) catalyst and solar
(pump A, 60 μL min− 1), binary capillary pump for carrying the mobile light for the degradation experiments.
phase (pump B), and variable wavelength UV-Vis detector with four The plates (240 × 445 mm) were cleansed first with detergent and
channels (10-mm optical path length, 45-nL cell volume). The analytical then with 11 mol L− 1 nitric acid solution. An aqueous TiO2 suspension
column was an Acclaim PepMap RSLC C18 (0.003 × 150 mm, 2 µm with a mass ratio of 1% at pH 3 (HNO3) was sprayed over the plates,
particle size). which were then oven-dried at 100ºC. This procedure was repeated 10
The six-port valve had two positions, A and B, as shown in Fig. 1. At times.
position A, 125 μL of sample were carried to the SPE-online column by The reactor was operated in recycle mode. The hormone-
pump A, while pump B conditioned the analytical column. After 4 min, contaminated water remained under magnetic stirring. The reactor
the valve was switched to position B, the sample was carried to the flow was controlled by a valve/rotameter. A bypass prevented pump
analytical column by pump B, while pump A cleansed the SPE column. malfunctioning. There was also a system to replace the water lost by
After 12 min of injection, the valve was switched back to position A, evaporation. Between samples, the reactor was completely drained.
conditioning the column for a new run. Afterwards, 2 L of tap water were allowed to circulate through the
The chromatographic runs were performed in reverse mode and reactor, under the sun, for at least 15 min.
gradient elution, using water (A) and acetonitrile (B): 0–4 min, 40% B; During the photocatalytic experiments, the incident solar radiation
4–6 min, 40–60% B; 6–10 min, 60% B; 10–11 min, 60–70% B; was continuously monitored (every 15 min) by a StellarNet EPP2000C
11–12 min, 70–40% B. The method qualification is detailed in the spectroradiometer at the same height and angle of the irradiated plates.

3
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 3. Preliminary experiments: (a) dark sorption on the reactor (plates without TiO2); (b) dark sorption on the surface of the catalytic plates (with TiO2); (c) direct
(solar) photolysis; and (d) solar photocatalysis. Experimental conditions: 250 μg L− 1 of each hormone, pH 4.0, and flow rate 100 mL min− 1.

2.5. Preliminary experiments point (pI), respectively [21]. Flow rate levels were 150 and
300 mL min− 1, based on preliminary experiments (not shown here). All
First, the feasibility of the proposed reactor for degrading the four experiments were randomly performed to avoid systematic errors.
hormones studied here was assessed. Moreover, three controls were Coding equations: x1 = pH−1.75
6.25
and x2 = Flow rate− 225
.
75
performed to be sure that the solar TiO2-photocatalytic degradation
would be significantly higher than direct photolysis and dark sorption.
Although tap water has a slightly acidic nature, pH was set to 4.0, trying 2.7. Estrogenic activity
to avoid major changes in water acidity during the experiments.
The initial concentration of each hormone was 250 µg L–1 and the Adherent MCF-7 cells were cultivated in Dubelco’s modified Eagle
reactor flow rate was arbitrarily set at 100 mL min− 1. The controls were: medium (DMEM) supplemented with glucose (3.5 g L− 1), 10% (v/v)
(a) dark sorption of the hormones in the reactor (plates without TiO2); fetal bovine serum (FBS), 1% (v/v) penicillin/streptomycin solution
(b) dark sorption of the hormones on TiO2; and (c) direct photolysis [22] in a 37 ◦ C incubator with 5% CO2 atmosphere and 90% humidity.
(plates without TiO2). For comparison purposes, (d) solar photocatalysis At 80% confluence, cells were washed with PBS and trypsin, incubated
at these very same conditions was performed. for 10 min, and centrifuged at 250 RCF (relative centrifugal force) for
5 min. The supernatant was removed, and the cells were resuspended
with medium to reach 1.0 × 105 cells mL− 1. Then, 100 μL were trans­
2.6. Factorial design ferred to a 96-well plate, following 24 h incubation. The medium was
replaced by charcoal-treated DMEM without phenol red with the same
After the significance of the solar TiO2-photocatalytic degradation supplements previously described. The preparation of the
was checked, the reactor performance was optimized using a 22 factorial charcoal-treated medium was described by Gomes et al. [19]. The pos­
design (two factors, two levels) in duplicate (Table 1). The studied itive control was made with 10 nmol L− 1 estradiol in 0.5% (v/v) ethanol,
factors were pH (x1 ) and flow rate (x2 ). while the negative control only had the treated medium with 0.5%
pH levels were set at 4.5 and 8.0. Those values are approximately a ethanol. The same procedure was performed with reactor samples
hundred times more acidic and more alkaline than the TiO2 isoelectric (containing the analytes), diluted 90 times. The positive control

4
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 4. Pareto charts for the results of the performed 22 full factorial design: (a) Estrone (E1), (b) 17β-estradiol (E2), (c) 17α-ethinylestradiol (EE2), and (d) Es­
triol (E3).

response at time 0 h was used as the reference for the estrogenic activity Limit of Detection (LoD) and the Limit of Quantification (LoQ) for the
of the water samples. four hormones were determined as 3.2 and 10 μg L− 1, respectively.
Diluted samples were sterilized by a polyvinylidene difluoride Linearity was tested from 10 to 250 μg L− 1, with R2 ranging from 0.987
(PVDF) 0.22 µm membrane and inoculated into the medium. Cells were to 0.997 (weighted linear regression model).
cultivated up to six days at 37ºC in an incubator, where the medium was The asymmetries of the chromatographic bands were 1.0–1.2 for all
replaced every 48 h. The estrogenic activity was measured by the MTT tested analytes; their respective chromatographic resolution varied from
colorimetric assay, which is based on the conversion of 3-(4,5-dime­ 2.34 to 3.03. Intraday precisions were estimated as the obtained relative
thylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, yellow) to standard deviations (RSD), which ranged from 0.758% to 14.0% for the
formazan (violet) by alive cells. The violet crystals were solubilized and four analytes; inter-day precisions varied from 4.5% to 17.6%. Accu­
the respective absorbances were measured at 570 nm [23]. racies were above 80% for three concentration levels (15, 130, and
230 µg L–1); similar figures were obtained for the inter columns accuracy
3. Results and discussion (10, 150, and 250 µg L–1). The matrix effect was measured by the re­
covery of the four analytes, which ranged from 80.9% to 90.4%. All
3.1. Chromatographic method qualification those values are considered acceptable by a series of guidelines [28].

Table 2 shows the structures and some properties of the four studied 3.2. Preliminary experiments
hormones. It is possible to observe that those four molecules have
similar structures and molecular masses, with pKa values showing that Fig. 3 shows the results obtained in the control experiments. One can
deprotonation occurs in alkaline pH (above 10), which is unlikely to observe (Fig. 3a) that even after 20 h of dark sorption (without TiO2
happen in the studied matrix (tap water). Those molecules are water- plate), hormone removal did not exceed 20%. Regarding dark sorption
soluble on a milligram-per-liter basis. logDpH=7.4 values suggest that on the TiO2 plate and direct (solar) photolysis (Fig. 3b and c), removals
those molecules interact with both hydrophilic and hydrophobic sub­ were approximately 5% and 10%, respectively. Those removals can be
stances [24]. Therefore, reverse-phase liquid chromatography is neglected when compared to the ones obtained in the photocatalytic
adequate for separation purposes. tests (more than 90% for all hormones in 3 h, Fig. 3d).
The qualification results of the chromatographic method are detailed
in the Supplementary Material (Figs. S1–S7, Tables S1–S6). In short, the

5
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 5. Flow rate optimization: (a) Estrone (E1), (b) 17β-estradiol (E2), (c) 17α-ethinylestradiol (EE2), and (d) Estriol (E3).

3.3. Optimization experiments However, those two pH were tested because they could be important for
the control experiments of the system (dark adsorption and direct
Table S7 shows the results of the 22 factorial design performed. The photolysis).
dependent variable (response) was the ratio between the percentual As pH was not statistically significant (within the tested range),
degradation of each hormone and the average incident energy at each further experiments were performed at the natural pH of the tap water
experiment. As those averages were different for each experimental used, i.e., approximately 6.8. Flow rate was optimized by varying it from
condition of the design, normalization of the data was required to allow 150 to 400 mL min− 1, at 50 mL min− 1 steps (the smallest division of the
for experiments to be compared. rotameter scale). The best flow rate range was between 200 and
Fig. 4 shows the respective Pareto chart for each response. In all of 250 mL min− 1 (Fig. 5). During the four hours of each experiment, flow
them, flow rate was the only statistically significant factor at 95% con­ rate decreased a little (approximately 20 mL min− 1), but in a repro­
fidence interval (bars that cross the red line), within tested levels. Flow ducible way. Therefore, it was set at 250 mL min− 1 at the beginning of
rate exhibited a positive value in all cases, meaning that to achieve the experiments so that, by their end flow rate was always within that
higher degradations, this factor must be increased. Up to a certain flow range.
rate, this is an expected result: as the degradation takes place only on the Fig. 5 shows a typical behavior regarding flow rate. As previously
plate, the higher the number of passes, the higher the degradation. pH described, as the number of passes increases, so does degradation.
was not statistically significant at 95% confidence interval. As the pKa of However, when exceedingly high flow rates are applied to the system,
all tested hormones are higher than 10, they had no charge at the tested the reduced contact time between the hormones and TiO2 impairs
pH (4.5 and 8.0); similarly, as the TiO2 pI (isoelectric point) is 6.25 [21], degradation.
the TiO2 surface was positive at pH 4.0 and negative at pH 8.0. There­
fore, it was highly expected that the pH factor showed no statistical
significance, as no electrostatic attraction occurred at both tested levels.

6
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 6. Degradation/(incident) energy ratio profile for the four hormones (in triplicate): (a) Estrone (E1), (b) 17β-estradiol (E2), (c) 17α-ethinylestradiol (EE2), and
(d) Estriol (E3).

3.4. Degradation/energy ratio profile ionization in the negative mode and few intense signals in the positive
one. In negative mode, the ionization is based on the deprotonation of
Once the parameters were optimized, the degradation/(incident) hydroxyl groups, which is favored by alkaline pH. As the reverse phase
energy ratio profiles could be obtained. That is an important measure as bonded to the stationary one has low tolerance to high pH values,
it is not possible to directly compare two degradation experiments 0.1 mL min–1 acetonitrile with 1% NH4OH was added post-column using
performed under different climatic conditions and/or different places. a Shimadzu LC-20 CE pump, before the ionization source.
Nevertheless, the degradation obtained with a certain dose of energy is Even with those higher analytical signals, it was not possible to
independent of those experimental conditions, therefore being useful for obtain tandem MS experiments. Therefore, to assess the obtained sig­
comparisons. Fig. 6 shows those profiles for the four hormones. nals, saturated solutions of individual hormones were degraded as well.
It is possible to observe that with 5.5 J cm− 2, more than 80% of the It is noteworthy that no degradation products were identified for 17α-
hormones were degraded. Degradations with more than that dose were ethinylestradiol (EE2) with the analytical conditions used. MZmine 2
difficult to measure because concentrations lower than the LoQ were freeware [29] was used for the alignment and deconvolution of signals.
reached. The degradation products identification was performed by comparing
As all experiments herein described were performed with a single the data with the literature.
TiO2-coated plate, and replicates were always consistent with one Two potential products were detected during the degradation of E1.
another, one can say that the TiO2-coated plate exhibited extended Product A, with m/z 303.1558 Da, was observed after approximately
reusability. 90 min. That mass-to-charge ratio corresponds (less than 16 ppm error)
to the structures proposed by Mazellier, Méité, and DeLaat [30]
3.5. Degradation products identification (Fig. 7a). Product B, with m/z 335.1447 Da, was detected since the early
stages of the degradation, reaching a maximum at 120 min. The prob­
During mass analyses, the four tested hormones exhibited low able structure for such degradation product is based on Ohko et al. [31]

7
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 7. Temporal profiles of the detected degradation products and their respective proposed structures: E1 (estrone) → (a) Product A (m/z 303.1558 Da) and (b)
Product B (m/z 335.1447 Da); E2 (17β-estradiol) → (c) Product C (m/z 335.1455 Da); and E3 (estriol) → (d) Product D (m/z 319.1544 Da).

Fig. 8. Hydroxylation mechanism of E1 to Product A or E3 to Product D.

(Fig. 7b). Regarding the degradation pathways, it is possible to say that


One potential product was detected in the E2 degradation: Product C, Product A is probably formed by the direct hydroxylation of E1; the
with m/z 335.1455 Da, which reached its maximum at approximately corresponding mechanism is proposed based on the work of Raghavan
120 min. The presented structure is also based on the work of Ohko et al. and Steeken [33] (Fig. 8). Those authors proposed that the hydroxyl
[31] (Fig. 7c). radical may attack the aromatic double bonds, adding itself to the ring.
One potential product was detected during the E3 degradation: Afterwards, radicalar hydrogen is eliminated to restore aromaticity.
Product D, with m/z 319.1544 Da, which was detected since the early Those radicals are mainly captured by hydroxyl radicals, generating
stages of the degradation and whose concentration increased thereafter. water as a secondary product. The same mechanism explains the for­
That structure is similar (errors below 5 ppm) to other ones proposed by mation of Product D, which is probably obtained by the direct hydrox­
Mazellier, Méité, and DeLaat [30] and Pereira et al. [32] (Fig. 7d). ylation of E3.

8
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

Fig. 9. Degradation mechanism of E1 to Product B.

Fig. 10. Estrogenic activity of samples and controls after six days of incubation (given as absorbance). Hormone degradation: (a) approximately 85%; (b) higher than
96% (below LoQ). Note: n = 6 (two independent experiments in triplicate). ANOVA analysis with Dunnet test is shown as *** p < 0.001 and **** p < 0.0001 in
relation to the 10 nmol L–1 estradiol (E2) positive control. Statistical analyses were performed with GraphPad Prism software.

Products B and C can be obtained from E1 and E2, respectively, after This behavior was expected because, at the end of the degradation ex­
the double hydroxylation of the aromatic ring and further oxidation, periments, approximately 15% (or 38 μg L− 1) of the initial hormone
resulting in the opening of the ring, as described by Hoffmann et al. [34] concentration remained in all samples. As Purdom et al. [7] state, hor­
(Fig. 9). In this mechanism, the hydroxyl groups can be oxidized to mone concentration as low as 1 ng L− 1 can still have estrogenic activity.
ketones. Regarding E2, only one of its hydroxyl groups underwent To assess the possibility of removing the estrogenic activity, long-
oxidation. Then, one hydroxyl radical and one dioxygen attacked the term degradation experiments were performed (9 h), shown in Fig. 9b.
hormone molecule adding a peroxide group to the ring. Finally, the Again, even after the degradation time was tripled, both the ANOVA
aromaticity was lost and the ring opened, generating a dicarboxylic acid, (Table S10) and Dunnet test (Table S11) showed that the estrogenic
with the elimination of a hydroperoxyl radical. activity could still be observed. However, in this extended period, the
estrogenic activity was reduced in comparison to the positive control
(Fig. 10b), pointing out that there was a reduction in the estrogenic
3.6. Estrogenic activity
activity, even though the hormone concentrations after 9 h of irradia­
tion could not be quantified once they were below the LoQ. It is known
The estrogenic activity test was performed using the positive controls
that the degradation was higher than 96% (LoQ/C0 = 0.04). Therefore,
(medium + estradiol) as the endpoint to be compared with the samples.
it was not possible to assess whether the degradation products them­
Positive and negative controls, as well as four samples from a 3-hour
selves caused estrogenic activity, due to the remaining concentration of
experiment (0, 1, 2, and 3 h) were tested in sextuplicate. Fig. 8a
the hormone.
shows the measured estrogenic activity, where absorbance and estro­
genicity are positively correlated.
4. Conclusions
The Analysis of Variance (one-way ANOVA, Table S8) shows that the
groups are not homogeneous. Then, the Dunnett multiple comparison
When the effects of two factors (pH and flow rate) on the degradation
tests (Table S9) were performed showing that the estrogenic activity of
of tap water contaminated with estrone, 17β-estradiol, 17α-ethinyles­
all samples was statistically the same (i.e., no estrogenic activity
tradiol, and estriol were assessed by a factorial design, pH was not
decrease could be noticed) except for the negative control (Fig. 10a).

9
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

statistically significant within the tested range (4.5 – 8.0). That finding [7] C.E. Purdom, P.A. Hardiman, V.V.J. Bye, N.C. Eno, C.R. Tyler, J.P. Sumpter,
Estrogenic effects of effluents from sewage treatment works, Chem. Ecol. 8 (4)
indicates the robustness of solar TiO2-photocatalysis for treating
(1994) 275–285.
hormone-laden tap water. [8] A. Mishra, A. Mehta, M. Sharma, S. Basu, Enhanced heterogeneous
After careful reactor optimization, solar (TiO2) photocatalysis could photodegradation of VOC and dye using microwave synthesized TiO2/Clay
remove high hormone concentrations (250 μg L–1) in a relatively short nanocomposites: a comparison study of different type of clays, J. Alloy. Compd.
694 (2017) 574–580.
period (4 h). Nevertheless, even extending degradation by three times, [9] R. Sanz, L. Romano, M. Zimbone, M.A. Buccheri, V. Scuderi, G. Impellizzeri,
the estrogenic activity of the treated tap water could not be eliminated. M. Scuderi, G. Nicotra, J. Jensen, V. Privitera, UV-black rutile TiO2: an
The automated, high-throughput chromatographic-UV-mass spec­ antireflective photocatalytic nanostructure, J. Appl. Phys. 117 (7) (2015), 074903.
[10] B.K. Mutumba, G.N. Shao, W.D. Kim, H.T. Kim, Sol–gel synthesis of mesoporous
trometry procedure herein employed was deemed suitable for quanti­ anatase–brookite and anatase–brookite–rutile TiO2 nanoparticles and their
fication purposes, regarding its calculated figures-of-merit. However, as photocatalytic properties, J. Colloid Interface Sci. 442 (2015) 1–7.
far as structural elucidation is concerned, it did not produce enough [11] X. Jiang, M. Manawan, T. Feng, R. Qian, T. Zhao, G. Zhou, F. Kong, Q. Wang,
S. Dai, J.H. Pan, Anatase and rutile in evonik aeroxide P25: heterojunctioned or
information about the formed fragments. Due to the nature of the hor­ individual nanoparticles? Catal. Today 300 (2018) 12–17.
mone molecules, electrospray sources produced incipient ionization. [12] Y.K. Abdel-Maksoud, E. Imam, A.R. Ramadan, Sand supported TiO2 photocatalyst
Therefore, the identification of structures was impaired, as the obtained in a tray photo-reactor for the removal of emerging contaminants in wastewater,
Catal. Today 313 (2018) 55–62.
instrumental signals were too low to allow for fragmenting the ions [13] J. Colina-Márquez, F. Machuca-Martínez, G.J.M. Li Puma, Modeling the
found. photocatalytic mineralization in water of commercial formulation of estrogens 17-
β estradiol (E2) and nomegestrol acetate in contraceptive pills in a solar powered
compound parabolic collector, Molecules 20 (7) (2015) 13354–13373.
CRediT authorship contribution statement [14] H.G. Oliveira, L.H. Ferreira, R. Bertazzoli, C. Longo, Remediation of 17-
α-ethinylestradiol aqueous solution by photocatalysis and electrochemically-
Rodrigo Nogueira Padovan: Methodology, Validation, Formal assisted photocatalysis using TiO2 and TiO2/WO3 electrodes irradiated by a solar
simulator, Water Res 72 (2015) 305–314.
analysis, Investigation, Writing – original draft. Lucas Sponton de
[15] D.H. Quiñones, P.M. Álvarez, A. Rey, F.J. Beltrán, Removal of emerging
Carvalho: Methodology, Validation, Formal analysis, Investigation. contaminants from municipal WWTP secondary effluents by solar photocatalytic
Patrícia Luisa de Souza Bergo: Investigation. Chubraider Xavier: ozonation. A pilot-scale study, Sep. Purif. Technol. 149 (2015) 132–139.
[16] M.C. Campos-Mañas, P. Plaza-Bolaños, J.A. Sánchez-Pérez, S. Malato, A. Agüera,
Formal analysis, Writing – original draft, Visualization. Andrei Leitão:
Fast determination of pesticides and other contaminants of emerging concern in
Resources, Writing – review & editing. Álvaro José dos Santos Neto: treated wastewater using direct injection coupled to highly sensitive ultra-high
Methodology, Resources, Writing – review & editing. Fernando Mauro performance liquid chromatography-tandem mass spectrometry, J. Chromatogr. A
Lanças: Resources. Eduardo Bessa Azevedo: Conceptualization, 1507 (2017) 84–94.
[17] T. Sultana, C. Murray, M.E. Hoque, C.D. Metcalfe, Monitoring contaminants of
Methodology, Resources, Data curation, Writing – review & editing, emerging concern from tertiary wastewater treatment plants using passive
Visualization, Supervision, Project administration. sampling modelled with performance reference compounds, Environ. Monit. Assess
(2017) 1–19.
[18] C. Ni, B. Zhu, N. Wang, M. Wang, S. Chen, J. Zhang, Y. Zhu, Simple column-
Declaration of Competing Interest switching ion chromatography method for determining eight monosaccharides and
oligosaccharides in honeydew and nectar, Food Chem. 194 (2016) 555–560.
[19] F.E.R. Gomes, P.L.S. Bergo, M.A. Trap, M. Spadoto, C.A. Galinaro, E. Rodrigues-
The authors declare that they have no known competing financial Filho, A. Leitão, G. Tremiliosi-Filho, Photolysis of parabens using medium-pressure
interests or personal relationships that could have appeared to influence mercury lamps: toxicity effects in MCF7, Balb/c 3T3 cells and Ceriodaphnia dubia,
the work reported in this paper. Chemosphere 208 (2018) 325–334.
[20] W.F.D. Vilela, A. Minillo, O. Rocha, E.M. Vieira, E.B. Azevedo, Degradation of [D-
Leu]-Microcystin-LR by solar heterogeneous photocatalysis (TiO2), Sol. Energy 86
Acknowledgement (9) (2012) 2746–2752.
[21] K. Suttiponparnit, J. Jiang, M. Sahu, S. Suvachittanont, T. Charinpanitkul,
P. Biswas, Role of surface area, primary particle size, and crystal phase on titanium
This study was funded by the São Paulo Research Foundation dioxide nanoparticle dispersion properties, Nanoscale Res. Lett. 6 (1) (2010)
(FAPESP, grants: 11/24137-0, 18/15904-6). 27–35.
[22] S. Comşa, A.M. Cîmpean, M. Raica, The story of MCF-7 breast cancer cell line: 40
years of experience in research, Anticancer Res. 35 (6) (2015) 3147–3154.
Appendix A. Supporting information [23] J.C. Stockert, R.W. Horobin, L.L. Colombo, A.J. Blázquez-Castro, Tetrazolium salts
and formazan products in Cell Biology: viability assessment, fluorescence imaging,
Supplementary data associated with this article can be found in the and labeling perspectives, Acta Histochem. 120 (3) (2018) 159–167.
[24] H. Cumming, C. Rücker, Octanol–water partition coefficient measurement by a
online version at doi:10.1016/j.jece.2021.106442. simple 1H NMR method, ACS Omega 2 (9) (2017) 6244–6249.
[25] K.M. Lewis, R.D. Archer, pKa values of estrone, 17β-estradiol and 2-
References methoxyestrone, Steroids 34 (5) (1979) 485–499.
[26] R. Marques, F.A. Vaz, H.C. Polonini, M.A. De Oliveira, Optimized separation
method for estriol, 17-β-Estradiol and progesterone by capillary
[1] The United Nations. World Water Development Report 2019: Leaving No One
electrochromatography with monolithic column and its application to a
Behind. Available on: https://en.unesco.org/themes/water-security/wwap/ww
transdermal emulsion, J. Braz. Chem. Soc. 26 (3) (2015) 609–618.
dr/. (Accessed 2 August 2020).
[27] M. Swain, chemicalize.org, J. Chem. Inf. Model. 52 (2) (2012) 613–615.
[2] G.T. Ankley, R.J. Erickson, D.J. Hoff, D.R. Mount, J. Lazorchak, J. Beaman, T.K.
[28] O. González, M.E. Blanco, G. Iriarte, L. Bartolomé, M.I. Maguregui, R.M. Alonso,
Linton, Aquatic life criteria for contaminants of emerging concern: Part I General
Bioanalytical chromatographic method validation according to current regulations,
challenges and recommendations. Available on: https://www.epa.gov/sites/
with a special focus on the non-well defined parameters limit of quantification,
production/files/2015-08/documents/white_paper/aquatic_life_criteria_for_
robustness and matrix effect, J. Chromatogr. A 1353 (2014) 10–27.
contaminants_of_emerging_concern_part_i_general_challenges_and_
[29] T. Pluskal, S. Castillo, A. Villar-Briones, M. Orešič, MZmine 2: modular framework
recommendations_1.pdf. (Accessed 2 August 2020).
for processing, visualizing, and analyzing mass spectrometry-based molecular
[3] A. Molinari, E. Sarti, N. Marchetti, L. Pasti, Degradation of emerging concern
profile data, BMC Bioinform. 11 (1) (2010) 395–406.
contaminants in water by heterogeneous photocatalysis with Na4W10O32, Appl.
[30] P. Mazellier, L. Méité, J.D. Laat, Photodegradation of the steroid hormones 17β-
Catal. B 203 (2017) 9–17.
estradiol (E2) and 17α-ethinylestradiol (EE2) in dilute aqueous solution,
[4] G. Lassonde, D. Nasuhoglu, J.F. Pan, B. Gaye, V. Yargeau, G. Delbes, Ozone
Chemosphere 73 (8) (2008) 1216–1223.
treatment prevents the toxicity of an environmental mixture of estrogens on rat
[31] Y. Ohko, K. Iuchi, C. Niwa, T. Tatsuma, T. Nakashima, T. Iguchi, Y. Kubota,
fetal testicular development, Reprod. Toxicol. 58 (2015) 85–92.
A. Fujishima, 17β-estradiol degradation by TiO2 photocatalysis as a means of
[5] S.X.L. Goh, H.K. Lee, An alternative perspective of hollow fiber-mediated
reducing estrogenic activity, Environ. Sci. Technol. 36 (19) (2002) 4175–4181.
extraction: Bundled hollow fiber array-liquid-phase microextraction with
sonication-assisted desorption and liquid chromatography–tandem mass
spectrometry for determination of estrogens in aqueous matrices, J. Chromatogr. A
1488 (2017) 26–36.
[6] S.Y. Wee, A.Z. Aris, F.M. Yusoff, S.M. Praveena, Tap water contamination:
multiclass endocrine disrupting compounds in different housing types in an urban
settlement, Chemosphere 264 (2021), 128488.

10
R.N. Padovan et al. Journal of Environmental Chemical Engineering 9 (2021) 106442

[32] R.O. Pereira, C. Postigo, M.L. De Alda, L.A. Daniel, D. Barceló, Removal of [34] E.H. Hoffmann, A. Tilgner, R. Wolke, O. Böge, A. Walter, H. Herrmann, Oxidation
estrogens through water disinfection processes and formation of by-products, of substituted aromatic hydrocarbons in the tropospheric aqueous phase: kinetic
Chemosphere 82 (6) (2011) 789–799. mechanism development and modelling, Phys. Chem. Chem. Phys. 20 (16) (2018)
[33] N.V. Raghavan, S. Steeken, Electrophilic reaction of the hydroxyl radical with 10960–10977.
phenol. Determination of the distribution of isomeric dihydroxycyclohexadienyl
radicals, Radic. J. Am. Chem. Soc. 102 (10) (1980) 3495–3499.

11

You might also like