Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
Contents lists available at SciVerse ScienceDirect
Journal of Photochemistry and Photobiology A: Chemistry journal homepage: www.elsevier.com/locate/jphotochem
Photocatalytic (UV-A/TiO2 ) degradation of 17␣-ethynylestradiol in environmental matrices: Experimental studies and artificial neural network modeling Zacharias Frontistis a , Vasileia M. Daskalaki a , Evroula Hapeshi b,c , Catherine Drosou a , Despo Fatta-Kassinos b,c , Nikolaos P. Xekoukoulotakis a , Dionissios Mantzavinos a,c,∗ a
Department of Environmental Engineering, Technical University of Crete, GR-73100 Chania, Greece Department of Civil and Environmental Engineering, University of Cyprus, 75 Kallipoleos St., 1678 Nicosia, Cyprus c NIREAS – International Water Research Center, University of Cyprus, Cyprus b
a r t i c l e
i n f o
Article history: Received 26 September 2011 Received in revised form 4 May 2012 Accepted 12 May 2012 Available online xxx Keywords: ANN Endocrine disrupting compounds Kinetics Synergy Transformation products Ultrasound
a b s t r a c t The efficiency of heterogeneous photocatalysis to degrade 17␣-ethynylestradiol (EE2), a synthetic estrogen hormone, in environmentally relevant samples was investigated. In most cases, UV-A radiation at a photon flux of 2.81 × 10−4 einstein/min was provided by a 9 W lamp and experiments were conducted at various concentrations of Aeroxide P25 TiO2 (50–1000 mg/L), EE2 concentrations (50–900 g/L) and water matrices (from ultrapure water to secondary treated wastewater). Some runs were performed at photon fluxes between 6.4 × 10−7 and 3.7 × 10−4 einstein/min to study the effect of intensity on degradation. Changes in estrogen concentration were followed by high performance liquid chromatography. EE2 degradation, which follows first order kinetics, increases with (i) increasing catalyst loading up to a threshold value beyond which it remains unaffected; (ii) increasing photon flux and (iii) decreasing matrix complexity, i.e. the organic and inorganic constituents of wastewater retard degradation. This may be overcome coupling photocatalysis with ultrasound radiation at 80 kHz and 41 W/L power density; the combined sonophotocatalytic process acts synergistically toward EE2 degradation. Several transformation products were identified by means of UPLC–MS/MS and a reaction network for the photocatalytic degradation of EE2 is suggested. An artificial neural network comprising five input variables (reaction time, TiO2 and EE2 concentration, organic content and conductivity of the water matrix), thirteen neurons and an output variable (EE2 conversion) was optimized, tested and validated for EE2 degradation. The network, based on tangent sigmoid and linear transfer functions for the hidden and input/output layers, respectively, and the Levenberg–Marquardt back propagation training algorithm, can successfully predict EE2 degradation. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Endocrine disrupting compounds (EDCs) constitute an important class of emerging environmental contaminants, which pose an increasing threat to aquatic organisms, as well as to human health. EDCs, which include natural estrogens, synthetic estrogens, phyto-estrogens and various industrial chemicals (i.e. pesticides, persistent organochlorines, organohalogens, alkyl phenols, heavy metals), have the ability to interact with the endocrine system of the organisms, thus leading to a variety of developmental
∗ Corresponding author at: Department of Environmental Engineering, Technical University of Crete, GR-73100 Chania, Greece. Tel.: +30 2821037797; fax: +30 2821037852. E-mail address:
[email protected] (D. Mantzavinos). 1010-6030/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jphotochem.2012.05.007
and reproductive disorders, as well as feminizing effects [1,2]. Of the various categories, natural and synthetic estrogens exhibit much stronger estrogenic activity than phyto- and xeno-estrogens [1]. EDCs are only partially removed in conventional wastewater treatment plants (WWTPs) mainly through a combination of biodegradation processes and sorption onto microbial flocs and suspended solids, although the relative contribution of each pathway is not fully understood [1]. Several monitoring campaigns [3–7] have confirmed the presence of estrogen hormones in WWTP discharges worldwide, including the naturally occurring estrone (E1) and 17-estradiol (E2), as well as 17␣-ethynylestradiol (EE2), a synthetic estrogen used in the oral contraceptive pill. These studies [1–7] and many more converge to the fact that WWTP discharges contain residual estrogens at the ng/L level that constitute the main contributors to the effluent’s estrogenic activity; in this respect,
Z. Frontistis et al. / Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
additional treatment may be needed to remove EDCs from the effluent. In recent years, advanced oxidation processes (AOPs) have gained particular attention for the degradation of microcontaminants of emerging concern in various aqueous matrices with photochemical and photocatalytic processes playing a key role [8]. Semiconductor photocatalysis based on TiO2 has been employed to degrade a mixture of E1 and E2 in model aqueous solutions concerning the effect of operating conditions (i.e. radiation wavelength, pH, catalyst and H2 O2 concentration) on kinetics [9]. In further studies [10], a synthetic solution of E1, E2, EE2 and estriol was subjected to TiO2 photocatalysis under UV-C or UV-A radiation in an annular photoreactor to model the radiation field and, subsequently, develop kinetic expressions independent of the reactor geometry and field. EE2 in pure water or in water:methanol mixture was photocatalytically oxidized to identify its major transformation intermediates and suggest possible degradation pathways [11]. The effect of adding ethanol and/or urea in aqueous solutions of E2 and EE2 on the extent of estrogen adsorption onto titania particles and subsequent photodegradation was evaluated in other studies [12]. A pilot system comprising a slurry photocatalytic reactor and a microfiltration membrane was tested to treat a mixture of 32 pharmaceuticals and EDCs (including five estrogen hormones) spiked in drinking water; emphasis was given on the effective removal of target compounds and associated estrogenicity in relation to energy consumption of the UV lamps [13]. Besides TiO2 photocatalysis [9–13], the use of iron oxide-coated resins for the heterogeneous photo-Fenton oxidation of E2 in distilled water has been reported [14,15]. Although the aforementioned studies [9–15] are important describing mostly experimental work in the area, they have exclusively been carried out in model aqueous solutions and/or at conditions that are far from those typically met in environmental samples (e.g. presence of organic solvents, estrogen concentrations at the mg/L level, highly acidic or alkaline conditions, etc.). In this perspective, the aim of our work was to study the UV-A/TiO2 degradation of EE2 in environmentally relevant samples (i.e. secondary treated effluents spiked with EE2 at the g/L level) with emphasis on the effect of water matrix, catalyst type and concentration, estrogen concentration, photon flux and ultrasound radiation on degradation kinetics. An attempt was also made to elucidate reaction pathways and mechanisms through the identification of primary transformation products (TPs). In addition to experimentation, artificial neural networks (ANNs) were employed to simulate the process and define the significance of the various operating variables. ANN are computerbased systems designed to simulate the learning process of neurons in the human brain [16]. In recent years, they have been employed in many areas of science and engineering to model various complex processes including biological and physicochemical water/wastewater treatments [17]. Notably, the number of publications dealing with ANN modeling in heterogeneous photocatalysis is rather limited as pointed out in a recent review article [16].
2. Experimental and analytical 2.1. Materials The majority of the experiments were performed with 17␣-ethynylestradiol (C20 H24 O2 ), which was purchased from Sigma–Aldrich (CAS number: 57-63-6) and used as received. Its structural formula and absorbance spectrum are shown in Schematic 1. Some runs were done with other EDCs, namely bisphenol A (BPA) and 17-estradiol, which were also purchased from Sigma–Aldrich and used as received.
0.4
Absorbance, au
34
CH OH C CH
0.3
H
0.2
H
HO 0.1
0.0 200
220
240
260
280
300
320
340
Wavelength, nm Scheme 1. The structural formula and absorbance spectrum of EE2.
Three commercially available titania samples were tested, namely: (i) Aeroxide P25 (formerly known as Degussa P25, 75:25 anatase:rutile, 50 m2 /g BET area, 21 nm particle size, supplied by Evonik Industries); (ii) Hombikat UV100 (>99% anatase, >250 m2 /g BET area, 5 nm particle size, supplied by Sachtleben Chemie); (iii) Tronox AK1 (100% anatase, 90 m2 /g BET area, 20 nm particle size, supplied by Tronox Inc.). Stock estrogen solutions were prepared at 1 mg/L concentration and the appropriate volume was spiked to the water matrix to obtain the desired estrogen concentration; four matrices were employed, namely: (i) wastewater (WW) collected from the outlet of the secondary treatment of a municipal WWTP. The matrix was characterized by standard methods [18] as follows: the chemical oxygen demand and dissolved organic carbon (DOC) was 24 and 8.4 mg/L, respectively, while its pH was about 8 and the conductivity 820 S/cm. Moreover, it contained 172 mg/L chlorides, 194 mg/L bicarbonates, 54 mg/L sulfates, 37 mg/L nitrates and 37 mg/L nitrites; (ii) ultrapure water (UPW) taken from a water purification system (EASYpureRF – Barnstead/Thermolyne, USA) with 5.5 S/cm conductivity and pH 6.1; (iii) a 50:50 mixture of WW and UPW; (iv) drinking water (DW) at pH 7.9, 308 S/cm conductivity and 152 mg/L bicarbonates. 2.2. Photocatalytic experiments Unless otherwise stated, UV-A radiation at 2.81 × 10−4 einstein/min was provided by a 9 W lamp (Radium Ralutec, 9W/78) emitting predominantly at 350–400 nm. The photon flux was determined actinometrically using the potassium ferrioxalate method. Experiments were conducted in an immersion well, batch type, laboratory scale photoreactor, purchased from Ace Glass (Vineland, NJ, USA) and described in detail elsewhere [19]. To study the effect of radiation intensity on degradation, experiments were also performed at various fluxes in the range of 6.4 × 10−7 –3.7 × 10−4 einstein/min; this was done either using a similar lamp of lower (i.e. 7 W) nominal intensity or partially covering the 9 W lamp with aluminum foil. In a typical run, 0.3 L of the water matrix containing EE2 in the range of 50–900 g/L were introduced in the reaction vessel made of borosilicate glass and then added titania in the range of 50–1000 mg/L. The suspension was magnetically stirred for 30 min in the dark and then the lamp was turned on, while air was continuously sparged in the reaction mixture under stirring. All experiments were performed at inherent pH which, although it was left uncontrolled during the reaction, remained practically unchanged. The temperature was maintained constant at 25 ± 2 ◦ C with a temperature control unit.
Z. Frontistis et al. / Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
2.3. Sonophotocatalytic experiments An Ultrason 250 (LabPlant, Huddersfield, UK) horn-type sonicator operating at 80 kHz was employed for sonophotocatalytic experiments, i.e. involving the simultaneous application of ultrasound and light radiation. Experiments were performed in an apparatus identical to that used in Section 2.2 but slightly modified to house both the UV-A lamp and the ultrasound probe. A glass tube at an angled position was connected to the external cylindrical reaction vessel near the surface of the liquid, thus permitting the introduction of the titanium-made probe inside the reaction mixture. The emitted ultrasound power was determined calorimetrically at 41 W/L. All other experimental procedures were similar to those described in Section 2.2. 2.4. Analytical protocols High performance liquid chromatography (HPLC: Alliance 2690, Waters) was employed to monitor the concentration of estrogens. Separation was achieved on a Luna C-18(2) column (5 m, 250 mm × 4.6 mm) and a security guard column (4 mm × 3 mm), both purchased from Phenomenex. The mobile phase consisting of 35:65 UPW:acetonitrile eluted isocratically at 1 mL/min and 30 ◦ C, while the injection volume was 100 L. Detection was achieved through a fluorescence detector (Waters 474 Scanning Fluorescence Detector) in which the excitation and emission wavelengths were set at 280 nm and 305 nm, respectively. At these conditions, the limits of detection and quantitation were 0.63 and 2.11 g/L for EE2, 0.68 and 2.32 g/L for BPA and 0.59 and 1.95 g/L for E2. TPs of EE2 photodegradation were identified by liquid chromatography coupled with mass spectrometry. Analysis was performed on an ACQUITY TQD UPLC–MS/MS system equipped with a sample manager, sample organizer, binary solvent manager and column manager. A triple quadrupole mass spectrometer (serial number QBA012) coupled with electrospray ionization (ESI) source running in positive mode was used for sample analysis. Data acquisition was performed with MassLynxTM software. In order to achieve sensitive analysis for the identification of TPs, data acquisition was performed with ESI in full scan mode. Separation was done on a BEH Shield RP18T column with the mobile phase consisting of UPW + 0.1% formic acid as eluent A and methanol as eluent B. The flow rate was 0.3 mL/min and the temperature 40 ◦ C. The elution gradient was: 0 min 5% B, 1.5 min 5% B, 2 min 30% B, 3 min 50% B, 5 min 70% B, 6 min 90% B, 7 min 90% B, 7.1 min 5% B, 9 min 5% B. The yeast estrogen screening (YES) test was employed to evaluate overall estrogenicity according to the procedures described elsewhere [20].
Table 1 Range and relative significance of the ANN input variables used in this work. Input variable
Range
Reaction time TiO2 concentration EE2 concentration Water matrix DOC Water matrix conductivity
0–60 min 50–1000 mg/L 50–900 g/L 0–8.4 mg/L 5.5–820 S/cm
Significance (%) 28 21.5 13.8 12.3 24.4 100
Total
conversion is the dependent variable of the output layer. A set of 222 experimental data was divided into training (70%, 156 data), validation (15%, 33 data) and test (15%, 33 data) subsets. The Neural Network Toolbox of Matlab R2011 mathematical software was employed for EE2 removal prediction. 4. Results and discussion 4.1. Preliminary experiments Preliminary runs were performed to screen three titania samples for EE2 degradation in WW; as seen in Fig. 1, photocatalytic activity decreases in the order Aeroxide P25 (23.1) > Hombikat UV100 (11) > Tronox AK1 (6.5) with the numbers in brackets showing initial (i.e. after 1 min of reaction) degradation rates in g/(L min). Although the surface area of Aeroxide P25 is lower than the other two, its superiority is attributed to the slower electron/hole recombination taking place on the catalyst surface compared to other TiO2 photocatalysts [21]. Another explanation ascribes the higher activity of Aeroxide P25 to its structure which is a mixture of anatase and rutile; this mixture is more active than the individual pure crystalline phases [22]. According to the above findings, all subsequent photocatalytic experiments were performed with Aeroxide P25 TiO2 . The degree of EE2 adsorption onto the catalyst surface in the dark was also monitored and it never exceeded 10% of its initial concentration irrespective of the conditions in question (i.e. type and concentration of titania, EE2 concentration and the water matrix). As also seen in Fig. 1, irradiation in the absence of catalyst yields just 25% conversion after 30 min which increases to 55% after 90 min (not shown); this confirms that EE2 degradation is predominantly due to the interaction between photonic energy and the catalyst surface and only partially due to photooxidation induced by the continuous oxygen supply in the liquid phase. At the conditions in question, direct photolysis is unlikely to occur since EE2 1
3. Artificial neural networks (ANN)
0.8
0.6
Tronox AK1 Hombikat P25 UV-A only
C/Co
A neural network consists of artificial neurons that are grouped into layers and interconnected in a variety of structures. The strength of these interconnections is determined by the weight associated with the neurons [16,17]. In this work, a three-layered back propagation ANN was chosen comprising an input layer (independent variables), an output layer (dependent variable) and a hidden layer. A tangent sigmoid (tansig) transfer function was employed to activate the hidden layer, while a linear (purelin) function for the input/output layers. The Levenberg–Marquardt back propagation algorithm was chosen for training purposes; more details regarding transfer functions and the training algorithm can be found elsewhere [16]. The input layer includes five variables which are shown in Table 1 alongside the respective range of values, while EE2
35
0.4
0.2
0 0
5
10
15
20
25
30
Time, min Fig. 1. Photodegradation of 100 g/L EE2 in WW with or without 750 mg/L of various TiO2 samples.
36
Z. Frontistis et al. / Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
70 UPW WW
60
ro, μg/(L.min)
50 40 30 20 10 0 0
200
400
600
800
1000
TiO2 concentration, mg/L
4.2. The role of water matrix
4.3. The effect of TiO2 concentration Fig. 3 shows the effect of increasing titania concentration on the initial photodegradation rate, ro , of EE2 in UPW and WW. As seen, the rate increases with increasing concentration up to 250–500 mg/L in UPW and 500–750 mg/L in WW beyond which it practically levels off. This threshold concentration (which is expectedly greater in more complex water matrices for the reasons discussed in Section 4.2) is characteristic of a heterogeneous
4.4. The effect of EE2 concentration Fig. 4a shows the effect of EE2 concentration on its photodegradation in WW at 750 mg/L Aeroxide P25 TiO2 ; regardless the initial 900
(a)
800 700
50μg/L 100μg/L 270μg/L 620μg/L 900μg/L
600 500 400 300 200 100 0 0
5
10
15
20
Time, min 0.4
(b) 200 0.35 150 0.3 100
0.25
ro, μg/(L.min)
The organic and inorganic species typically found in environmentally relevant matrices may interfere with the oxidizing agents (e.g. photogenerated holes, as well as reactive oxygen species including predominantly, but not exclusively, hydroxyl radicals) and affect the rate of substrate elimination. This is evident in Fig. 2 showing that the matrix has an adverse effect on conversion and/or reaction rate; for instance, the time needed for complete EE2 degradation in WW is three times longer than that in UPW and twice longer that that in WW diluted with an equal volume of UPW. This can be explained taking into account that (i) the oxidizing agents are competitively consumed in reactions involving the residual organic fraction present in treated WW but not in UPW. Since this is known to be refractory to biological or chemical oxidation [23] and constitutes most of the matrix’s total organic content (i.e. 94–99% depending on EE2 initial concentration), non-selective oxidizing agents will partly be wasted attacking this fraction; (ii) hydroxyl radicals may be scavenged by bicarbonates, chlorides and sulfates present in WW and/or DW to form the respective radicals, whose oxidation potential is lower than that of hydroxyl radicals [23]. The fact that the various matrices have pH values between ∼6 for UPW and ∼8 for WW and DW is not expected to alter the ionization state of either EE2 or the catalyst surface and, consequently, determine their interaction; this is so because EE2 with a pKa value of 10.2 [23] will exist in its molecular form in either matrix, while Aeroxide P25 TiO2 with a zero point charge of 6.2 [24] will be negatively charged. The effect of changing matrix pH on kinetics was not investigated in this work because this would not be practical in environmentally relevant samples such as WWTP discharges or groundwater destined for the production of drinking water.
catalytic regime since the fraction of incident light absorbed by the semiconductor progressively increases in suspensions containing higher amounts of TiO2 up to the point where all particles are fully illuminated [25].
Concentration, μg/L
does not absorb over 300 nm (as clearly seen in Schematic 1), which coincides with the transmittance threshold of borosilicate.
Fig. 3. Effect of Aeroxide P25 TiO2 concentration on the initial photodegradation rate (i.e. calculated after 1 min of reaction) of 100 g/L EE2.
kPC, min-1
Fig. 2. Effect of water matrix on 100 g/L EE2 photodegradation with 750 mg/L Aeroxide P25 TiO2 . Inset graph shows initial photodegradation rates (i.e. calculated after 1 min of reaction).
50
0.2
0 0
150
300
450
600
750
900
Concentration, μg/L Fig. 4. Effect of initial EE2 concentration on (a) concentration–time profiles; (b) rate constants and initial rates (calculated after 1 min of reaction) during photodegradation in WW with 750 mg/L Aeroxide P25 TiO2 .
Z. Frontistis et al. / Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
37
estrogen concentration, ≥95% conversion is always achieved after 10 min of reaction. Given that (i) the range of EE2 concentrations employed in this work is relatively low; (ii) the formation rate of the various oxidizing agents should be constant at fixed experimental conditions (i.e. photon flux, catalyst concentration, matrix), one could safely assume first order kinetics regarding EE2 concentration, i.e.: dC Co = kPC C ⇔ ln = kPC t C dt
(1)
where kPC is an apparent rate constant incorporating the (nearly) constant concentration of oxidizing agents. If the results of Fig. 4a up to 90% conversion are plotted in the form of Eq. (1), straight lines passing through the origin (not shown) fit the experimental data well with the coefficient of linear regression, R2 , being ≥99%. From the slopes of the resulting lines, kPC values can be computed and these are shown in Fig. 4b. For concentrations between 50 and 620 g/L, kPC is equal to 32.7 ± 0.6 × 10−2 min−1 , thus denoting true first order kinetics since the rate constant is independent of the estrogen concentration. Only at the higher concentration of 900 g/L, does kPC slightly decrease to 28 × 10−2 min−1 , which implies deviation from first order kinetics (although fitting to Eq. (1) is still excellent). This deviation can better be seen comparing reaction rates rather than rate constants. An increase in EE2 concentration increases the chance of oxidizing agents to attack the substrate, thus increasing the reaction rate; this is particularly true during the early stages when the concentration of competing transformation by-products is low. As seen in Fig. 4b, the relationship between concentration and ro is indeed linear in the range of 50–620 g/L. If this were the case at 900 g/L too, one could predict an initial rate of 225 g/(L min) extrapolating the respective line (marked –x– in Fig. 4b); nonetheless, the actual value is 12% lower indicating a transition from first to lower order kinetics. Previous studies dealing with the photocatalytic degradation of estrogen hormones have also reported first order kinetics [9,10,13]. In particular, Zhang et al. [9], who studied the degradation of E1 and E2 in the range of 100–1000 ng/L at 253 nm, reported rate constants equal to 2.42 ± 0.08 and 2.29 ± 0.02 h−1 for E1 and E2, respectively. When the experiments were repeated with a light source emitting in the range of 238–579 nm, the corresponding rate constants decreased by about 65%. 4.5. The effect of radiation intensity Fig. 5 shows the effect of altering photon flux in the range 6.4 × 10−7 –3.7 × 10−4 einstein/min on 100 g/L EE2 degradation in UPW. Complete EE2 removal occurs rapidly (i.e. within 1–3 min) at photon fluxes in the order of 10−4 einstein/min; nonetheless, a decrease in photon flux by 2–3 orders of magnitude results in a significant drop of photocatalytic performance, thus verifying the light-induced nature of the activation of the catalytic process. For example, the initial degradation rate suffers a 2.5-fold and 18-fold reduction, when the flux decreases from 3.7 × 10−4 to 4.6 × 10−6 and 6.4 × 10−7 einstein/min, respectively. 4.6. Reactivity of EDCs To assess the ability of photocatalysis to degrade other EDCs in WW, experiments were carried out with BPA and E2 and the results are shown in Fig. 6. The initial degradation rate is common for all three estrogens at 20 ± 3 g/(L min), while 15 min of reaction suffice to remove 95% of BPA and E2. These results are consistent with a recent work of our group [26] reporting comparable degradation rates for EE2 and BPA subject to solar light-induced photocatalysis over various titania samples in WW. Likewise, Li Puma et al. [10]
Fig. 5. Effect of photon flux on 100 g/L EE2 photodegradation in UPW with 250 mg/L Aeroxide P25 TiO2 . Inset graph shows initial photodegradation rates (i.e. calculated after 1 min of reaction).
reported similar rate constants for the degradation of E2 and EE2 under UV-A or UV-C radiation over Aeroxide P25 TiO2 . 4.7. Enhancement of photocatalytic performance by ultrasound radiation Coupling photocatalysis to other AOPs is a way to improve process efficiency due to cumulative or synergistic effects. The application of ultrasound (US) in an aqueous medium leads to the cyclic formation, growth and adiabatic collapse of cavities that behave as “hot spot” micro-reactors. There are two distinct mechanisms for organic matter decomposition involving (i) pyrolysis near or inside the cavity; (ii) reactions mediated by hydroxyl radicals whose formation is due to water sonolysis [27]. Recent studies have demonstrated the ability of US radiation to degrade various estrogen hormones [23,28,29], while others [27,30] have reported the beneficial effect of coupling US and photocatalysis for the mineralization of BPA. In this perspective, we decided to test EE2 degradation by US radiation, photocatalysis and their combination, i.e. sonophotocatalysis. As seen in Fig. 7, US alone results in partial EE2 elimination (e.g. 20% after 20 min) which is comparable to that of photooxidation. On the other hand, sonophotocatalysis leads to 90% conversion in just 4 min, twice as fast as photocatalysis. If the data of Fig. 7 up to 90% conversion are fitted to Eq. (1), one can compute the rate constants of sonophotocatalysis 1 BPA E2 EE2
0.8
0.6
C/Co
−
0.4
0.2
0 0
5
10
15
20
Time, min Fig. 6. Reactivity of 100 g/L EE2, E2 and BPA during photodegradation in WW with 750 mg/L Aeroxide P25 TiO2 .
38
Z. Frontistis et al. / Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
1
100
100
C/Co
0.6
60
UV-A US UV-A/US Photocatalysis Sonophotocatalysis YES
0.4
0.2
40
20
0
Estrogenicity, μg/L
80
Relative abundance, %
0.8
(a)
5
10
15
60
40
20
0
0 0
Peak 1
80
0
20
1
2
3
4
Time, min
(kUSPC = 60 × 10−2 min−1 ), photocatalysis (kPC = 32.3 × 10−2 min−1 ) and US radiation (kUS = 1.1 × 10−2 min−1 ) and then assess the degree of synergy, S, as follows: kUSPC − (kPC + kUS ) × 100 kUSPC
6
7
8
9
10
(2)
In this case, S = 44.3% indicating that the individual processes do not simply add up (i.e. S = 0) but there is a two-way synergy between them. On one hand, application of an ultrasound field can result in (i) enhanced mass transfer of EE2 between the liquid phase and the titania surface; (ii) catalyst excitation by sonoluminescence which has a wide wavelength range below 375 nm; (iii) increased catalytic activity due to ultrasound de-aggregating catalyst particles, thus increasing surface area [27]. On the other hand, the sonochemical activity itself may be facilitated by the presence of titania particles acting as nucleation sites (this said, particles may be responsible for sound attenuation leading to reduced degradation). Such synergistic effects could conceptually be crucial for the successful implementation of AOPs in water/wastewater treatment. A proper strategy is the simultaneous application of AOPs that could complement each other in a synergistic rather than additive way. Since the organic and inorganic, non-target constituents present in complex water matrices behave antagonistically to target species consuming a fraction of non-selective radicals, this could be compensated by the increased generation of radicals arising from process integration. This is clear from the profile of overall estrogenicity also shown in Fig. 7 and the way it changes during sonophotocatalysis. The WW sample contains 103 g/L of equivalent estrogenicity, 22 g/L of which are due to EE2 and the rest to matrix constituents. It is notable that the quantitative EE2 removal after 5 min of reaction coincides with 27% estrogenicity reduction, which remains nearly constant thereafter (i.e. at 30% after 20–30 min). This reduction is marginally greater than the contribution of EE2 to initial estrogenicity, implying that other resistant matrix species retain their estrogenic properties.
400
450
500
550
4.8. Identification of TPs and proposed pathways Several samples were analyzed by UPLC–MS/MS in an attempt to determine major, early-stage EE2 TPs. Fig. 8a shows a typical total ion chromatogram (TIC) of the mass spectra obtained from photocatalytically treated EE2 samples in UPW, where a major peak at 5 min (referred to as peak 1) can clearly be seen. Three main
100
(b) Relative abundance, %
Fig. 7. Degradation of 100 g/L EE2 in WW by 80 kHz ultrasound radiation, photocatalysis (750 mg/L Aeroxide P25 TiO2 ) and sonophotocatalysis. Changes in estrogenicity during sonophotocatalysis are shown in secondary axis.
%S =
5
Time, min 329.3
80
327.3
60 282.5 251.2
40
207.3
20
229.4
279.5
300.7
0 50
100
150
200
250
300
350
m/z Fig. 8. (a) TIC of mass spectra for EE2 photodegradation; (b) average mass spectra corresponding to peak 1 in TIC.
product ions (whose relative intensity (RI) is greater than 50%) at m/z 329, m/z 282, m/z 251, and four other important product ions (whose RI is between 20 and 50%) at m/z 301, m/z 279, m/z 229, m/z 207 are observed in the average mass spectra of peak 1 (Fig. 8b). Overall, as many as ten compounds were tentatively identified with their peaks being greater than 20% of the highest peak intensity, thus allowing the precise identification of TPs; these are shown in Schematic 2 (TP1–TP10), alongside the proposed reaction pathways for EE2 photocatalytic degradation in UPW. Based on the results obtained herein and on previous studies regarding EE2 photooxidation [31], TPs correspond to quinone methide and 1,2-quinone derivatives, ring-monohydroxylation photoproducts and photoproducts arising from the hydroxylation onto the saturated ring linked to the aromatic one. TP1(A) with 329 m/z (RI = 80–100%) is one of the major products, which may be formed after di-hydroxylation of the aromatic ring. The attack of hydroxyl radicals on the aromatic ring can lead to the formation of dihydroxy photoproducts such as 2-HO-EE2 [32]. In addition, the formation of TP1(B) may be due to the direct attack of OOH radicals [33,34]. As Peiro et al. reported [35], with the rearrangement of the structure at the phenol moiety, the resonance structure can be formed, where a site between aromatic and cyclohexane ring is active by the attack of OOH radical. TP2 (m/z 327, RI = 70–100%) can be formed from TP1(B) with the abstraction of two hydrogen atoms. EE2 dealkylation (i.e. loss of C2 H2 moiety) at position C17 could lead to the formation of its analogue, TP3, (m/z 272, RI = 20–70%), whose further reduction would yield TP4 (RI = 20–40%). Another analogue, TP5 (RI = 30–45%), may be formed following EE2 demethylation at position C13. Cleavage of the aromatic ring of EE2 also occurs
Z. Frontistis et al. / Journal of Photochemistry and Photobiology A: Chemistry 240 (2012) 33–41
39
Scheme 2. The proposed reaction network for the photodegradation of EE2 in UPW.
resulting in the formation of TP6 (m/z 251, RI = 40–70%), whose subsequent decarboxylation gives TP7 (m/z 207, RI = 20–35%). TP8 with m/z 279 (RI = 20–40%) may be formed from EE2 after losing a hydroxyl moiety from position 20, while TP9 with m/z 313 (RI = 25–50%) may be generated by direct attack of hydroxyl radical at position C10. Finally, TP10 with m/z 300 has been identified (RI = 40–60%) possibly coming from the addition of four hydrogen atoms to EE2. It should be noted here that the analytical efforts deliberately focused on the determination of early-stage intermediates rather than deep oxidation TPs since the former could give a better insight with respect to reaction mechanisms and pathways; in this respect,
EE2 conversion in any of the analyzed samples did not exceed 60%. Nonetheless, compounds with molecular mass TiO2 concentration > estrogen concentration ≈ matrix DOC.
[14]
[15]
[16]
[17]
[18] [19]
[20]
[21]
[22]
[23]
Acknowledgments This work was partly co-funded by the Republic of Cyprus and the European Regional Development Fund through Grants UPGRADING/DURABLE/0308/07 (Project title: IX-Aqua – Fate, effect and removal potential of xenobiotics present in aqueous matrices) and “NIREAS – International Water Research Center” (NEA IPODOMI/STRATH II/0308/09).
[24]
[25]
[26]
References [27] [1] M. Auriol, Y. Filali-Meknassi, R.D. Tyagi, C.D. Adams, R.Y. Surampalli, Endocrine disrupting compounds removal from wastewater: a new challenge, Process Biochemistry 41 (2006) 525–539. [2] M. Kuster, D.A. Azevedo, M.J. Lopez de Alda, F.R. Aquino Neto, D. Barcelo, Analysis of phytoestrogens, progestogens and estrogens in environmental waters from Rio de Janeiro (Brazil), Environment International 35 (2009) 997–1003. [3] M.R. Servos, D.T. Bennie, B.K. Burnison, A. Jurkovic, R. McInnis, T. Neheli, A. Schnell, P. Seto, S.A. Smyth, T.A. Ternes, Distribution of estrogens, 17-estradiol and estrone, in Canadian municipal wastewater treatment plants, Science of the Total Environment 336 (2005) 155–170. [4] C. Desbrow, E.J. Routledge, G.C. Brighty, J.P. Sumpter, M. Waldock, Identification of estrogenic chemicals in STW effluent. 1: chemical fractionation and in vitro biological screening, Environmental Science and Technology 32 (1998) 1549–1558. [5] T.A. Ternes, M. Stumpf, J. Mueller, K. Haberer, R.D. Wilken, M. Servos, Behavior and occurrence of estrogens in municipal sewage treatment plants. I: investigations in Germany, Canada and Brazil, Science of the Total Environment 225 (1999) 81–90. [6] C. Baronti, R. Curini, G. D’Ascenzo, A. Di Corcia, A. Gentili, R. Samperi, Monitoring natural and synthetic estrogens at activated sludge sewage treatment plants and in a receiving river water, Environmental Science and Technology 34 (2000) 5059–5066. [7] S. Matsui, H. Takigami, T. Matsuda, N. Taniguchi, J. Adachi, H. Kawami, Y. Shimizu, Estrogen and estrogen mimics contamination in water and the role of sewage treatment, Water Science and Technology 42 (2000) 173–179. [8] M. Klavarioti, D. Mantzavinos, D. Kassinos, Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes, Environment International 35 (2009) 402–417. [9] Y. Zhang, J.L. Zhou, B. Ning, Photodegradation of estrone and 17b-estradiol in water, Water Research 41 (2007) 19–26. [10] G. Li Puma, V. Puddu, H.K. Tsang, A. Gora, B. Toepfer, Photocatalytic oxidation of multicomponent mixtures of estrogens (estrone (E1), 17-estradiol (E2), 17␣-ethynylestradiol (EE2) and estriol (E3)) under UVA and UVC radiation: photon absorption, quantum yields and rate constants independent of photon absorption, Applied Catalysis B: Environmental 99 (2010) 388–397. [11] W. Sun, S. Li, J. Mai, J. Ni, Initial photocatalytic degradation intermediates/pathways of 17␣-ethynylestradiol: effect of pH and methanol, Chemosphere 81 (2010) 92–99. [12] T. Karpova, S. Preis, J. Kallas, Selective photocatalytic oxidation of steroid estrogens in water treatment: urea as co-pollutant, Journal of Hazardous Materials 146 (2007) 465–471. [13] M.J. Benotti, B.D. Stanford, E.C. Wert, S.A. Snyder, Evaluation of a photocatalytic reactor membrane pilot system for the removal of pharmaceuticals
[28]
[29]
[30]
[31]
[32]
[33] [34]
[35]
[36]
[37]
[38]
[39]
[40]
41
and endocrine disrupting compounds from water, Water Research 43 (2009) 1513–1522. Y. Zhao, J. Hu, Photo-Fenton degradation of 17-estradiol in presence of ␣-FeOOHR and H2 O2 , Applied Catalysis B: Environmental 78 (2008) 250–258. Y. Zhao, J. Hu, H. Chen, Elimination of estrogen and its estrogenicity by heterogeneous photo-Fenton catalyst -FeOOH/resin, Journal of Photochemistry and Photobiology A: Chemistry 212 (2010) 94–100. A.R. Khataee, M.B. Kasiri, Artificial neural networks modeling of contaminated water treatment processes by homogeneous and heterogeneous nanocatalysis, Journal of Molecular Catalysis A: Chemical 331 (2010) 86–100. E.S. Elmolla, M. Chaudhuri, M.M. Eltoukhy, The use of artificial neural network (ANN) for modeling of COD removal from antibiotic aqueous solution by the Fenton process, Journal of Hazardous Materials 179 (2010) 127–134. L.S. Clesceri, A.E. Greenberg, A.D. Eaton, Standard Methods for the Examination of Water and Wastewater, 20th ed., APHA, 1999. C. Fotiadis, N.P. Xekoukoulotakis, D. Mantzavinos, Photocatalytic treatment of wastewater from cottonseed processing: effect of operating conditions, aerobic biodegradability and ecotoxicity, Catalysis Today 124 (2007) 247–253. E.J. Routledge, J.P. Sumpter, Estrogenic activity of surfactants and some of their degradation products assessed using a recombinant yeast screen, Environmental Toxicology and Chemistry 15 (1996) 241–248. S.T. Martin, H. Herrmann, W. Choi, M.R. Hoffmann, Time-resolved microwave conductivity. Part 1: TiO2 photoreactivity and size quantization, Journal of the Chemical Society, Faraday Transactions 21 (1994) 3315–3322. R.I. Bickley, T. Gonzalez-Carreno, J.S. Lees, L. Palmisano, R.J.D. Tilley, A structural investigation of titanium dioxide photocatalysts, Journal of Solid State Chemistry 92 (1991) 178–190. Z. Frontistis, D. Mantzavinos, Sonodegradation of 17␣-ethynylestradiol in environmentally relevant matrices: Laboratory-scale kinetic studies, Ultrasonics Sonochemistry 19 (2012) 77–84. M. Saquib, M. Muneer, Titanium dioxide mediated photocatalyzed degradation of a textile dye derivative, acid orange 8, in aqueous suspensions, Desalination 155 (2003) 255–263. M.N. Abellan, J. Gimenez, S. Esplugas, Photocatalytic degradation of antibiotics: the case of sulfamethoxazole and trimethoprim, Catalysis Today 144 (2009) 131–136. H. Dimitroula, V.M. Daskalaki, Z. Frontistis, D.I. Kondarides, P. Panagiotopoulou, N.P. Xekoukoulotakis, D. Mantzavinos, Solar photocatalysis for the abatement of emerging micro-contaminants in wastewater: synthesis, characterization and testing of various TiO2 samples, Applied Catalysis B: Environmental 117–118 (2012) 283–291. R.A. Torres-Palma, J.I. Nieto, E. Combet, C. Petrier, C. Pulgarin, An innovative ultrasound, Fe2+ and TiO2 photoassisted process for bisphenol A mineralization, Water Research 44 (2010) 2245–2252. H. Fu, R.P.S. Suri, R.F. Chimchirian, E. Helmig, R. Constable, Ultrasound-induced destruction of low levels of estrogen hormones in aqueous solutions, Environmental Science and Technology 41 (2007) 5869–5874. R.P.S. Suri, T.S. Singh, S. Abburi, Influence of alkalinity and salinity on the sonochemical degradation of estrogen hormones in aqueous solution, Environmental Science and Technology 44 (2010) 1373–1379. R.A. Torres, J.I. Nieto, E. Combet, C. Petrier, C. Pulgarin, Influence of TiO2 concentration on the synergistic effect between photocatalysis and high-frequency ultrasound for organic pollutant mineralization in water, Applied Catalysis B: Environmental 80 (2008) 168–175. P. Mazelier, L. Méité, J. De Laat, Photodegradation of the steroid hormones 17b-estradiol (E2) and 17a-ethinylestradiol (EE2) in dilute aqueous solution, Chemosphere 73 (2008) 1216–1223. Y. Zhao, J. Hu, W. Jin, Transformation of oxidation products and reduction of estrogenic activity of 17-estradiol by heterogeneous photo-Fenton reaction, Environmental Science and Technology 42 (2008) 5277–5284. J. Mai, W. Sun, L. Xiong, J. Ni, Titanium dioxide mediated photocatalytic degradation of 17-estradiol in aqueous solution, Chemosphere 73 (2008) 600–606. B.E. Segmuller, B.L. Armstrong, R. Dunphy, A.R. Oyler, Identification of autoxidation and photodegradation products of ethynylestradiol by on-line HPLC–NMR and HPLC–MS, Journal of Pharmaceutical and Biomedical Analysis 23 (2000) 927–937. A.M. Peiro, J.A. Ayllon, J. Peral, X. Domenech, TiO2 -photocatalyzed degradation of phenol and ortho-substituted phenolic compounds, Applied Catalysis B: Environmental 30 (2001) 359–373. C.A. Emilio, M.I. Litter, J.F. Magallanes, Semiempirical modeling with application of artificial neural networks for the photocatalytic reaction of ethylenediaminetetraacetic acid (EDTA) over titanium oxide (TiO2 ), Helvetica Chimica Acta 84 (2001) 799–813. C.A. Emilio, J.F. Magallanes, M.I. Litter, Chemometric study on the TiO2 photocatalytic degradation of nitrilotriacetic acid, Analytica Chimica Acta 595 (2007) 89–97. A.R. Khataee, Photocatalytic removal of C.I. basic red 46 on immobilized TiO2 nanoparticles: artificial neural network modelling, Environmental Technology 30 (2009) 1155–1168. F.L. Toma, S. Guessasma, D. Klein, G. Montavon, G. Bertrand, C. Coddet, Neural computation to predict TiO2 photocatalytic efficiency for nitrogen oxides removal, Journal of Photochemistry and Photobiology A: Chemistry 165 (2004) 91–96. G.D. Garson, Interpreting neural-network connection weights, AI Expert 6 (1991) 46–51.
