CHARACTERIZATION-PERFORMANCE OF ZnO AND ZnO/ZnFe 2 O 4 CATALYST USING ARTIFICIAL AND SOLAR LIGHT FOR MERCURY (II) REDUCTION

- This study describes the application of ZnO commercial oxide and ZnO/ZnFe 2 O 4 mixed oxides to reduce the mercury ion (Hg 2+ ) in the aqueous phase. The catalysts synthesized were characterized by thermogravimetric analysis, adsorption and desorption of N 2 , scanning electron microscopy, X-ray diffraction, Fourier transform infrared spectroscopy, photoacoustic spectroscopy and Mössbauer spectroscopy. The heterogeneous photocatalysis process was studied in synthetic effluent treatment containing mercury chloride (HgCl 2 ) in a batch and solar reactor. The results indicated that commercial ZnO catalyst totally reduced the Hg (II), in all conditions and reaction systems studied. On the other hand, for the mixed oxides (ZnO/ZnFe 2 O 4 ), there was a decrease in the reduction efficiency in the batch reactor, indicating no ability to reduce Hg(II) in the presence of sunlight.


INTRODUCTION
Nowadays, one of the main environmental problems is the increasing contamination of water bodies by an excess of heavy metals, such as mercury (Hg), which is not only an environmental hazard, but also a serious public health problem (Zhou et al., 2017;Mudasir et al., 2016;Botta et al.2002).
Once present in water bodies, Hg is mainly in the form of HgCl 2 . Particles containing Hg 2+ can also be methylated by the action of specific bacteria, such as Desulfovibrio desulfuricans ND132 and Geobacter sulfurreducens PCA, resulting in monomethylmercury (MMHg, CH 3 Hg + ) and dimethylmercury (DMHg,CH 3 HgCH 3 ), which are considered the most toxic forms of Hg (Pieve et al., 2014).
In this context, heterogeneous photocatalysis is a promising environmental decontamination method, since it allows not only the removal of recalcitrant substances, but also their complete mineralization, avoiding the deposition of these compounds in the receiving aquatic bodies (Lenzi et al., 2011).
Several photocatalytic studies employ the use of titanium dioxide (TiO 2 ) as a semiconductor. However, zinc oxide (ZnO) has a considerably higher electron mobility (200-300 cm 2 V -1 s -1 ) than TiO 2 (0.1-4 cm 2 V -1 s -1 ), which favors a greater efficiency in the transfer of the electrons. In addition, the ZnO valence band is positioned slightly below the corresponding TiO 2 band, so the hydroxyl radicals generated by ZnO (+3.06 V) have a higher oxidation potential when compared to those generated by TiO 2 (+2.7 V) (Kumar et al., 2015;Mohamed et al., 2014). Therefore, ZnO absorbs more of the solar spectrum than other semiconductors, and it is considered an ideal material to be used in the presence of solar energy (Kumar et al., 2015;Mohamed et al., 2014;Lee et al., 2016).
Zinc ferrite (ZnFe 2 O 4 ) presents great potential for application in heterogeneous photocatalysis processes since it is low cost and has a stable structure. Besides, it presents a band gap of 1.9 eV, which makes it a promising material for photocatalytic applications under sunlight.
The recombination of the generated charges (e -/h + ) constitutes the major limitation of the heterogeneous photocatalysis process, because when the recombination occurs, the electron returns to its valence band without reacting with the species adsorbed on the surface of the semiconductor, dissipating the energy in the form of heat (Pelaez et al., 2012;Rengaraj et al., 2007). Several strategies have been adopted with the purpose of increasing the photocatalytic efficiency. These can be changes in the semiconductor structure and morphology, such as the synthesis of heteroconjugate structures and the distribution of the charges on the surfaces of the photocatalyst, or chemical modifications in the reaction medium, by adding an agent of sacrifice, for example, oxygen, methanol, ethanol, formic acid, oxalic acid, humic acid and others (Wang et al., 2013;Jang et al., 2012;López-Muñoz et al., 2011).
Mercuric species in solution cannot be bio-or chemically degraded (Bota et al., 2002). However, the possibility of the sorbent being reduced or not is very dependent on its standard reduction potential after the adsorption process (Rahman, et al., 2016). pH determines the specification of metal ions and therefore affects the ability of Hg (II) to be adsorbed (Rahman, et al., 2016) onto the ZnO/ZnFe 2 O 4 materials.
The microstructural properties of the photocatalyst particles are specific and related to the synthesis method, for this reason different techniques are used, such as precipitation, microemulsion, thermal decomposition (Music et al., 2002), ion exchange, chemical vapor deposition (Wang et al., 2013), solgel (Lenzi et al., 2011), among others. However, a considerable part of such methodologies is associated with high production costs, which make its application inappropriate for developing countries (Mudasir et al., 2016). In this context, impregnation with excess solvent is a promising technique since it is simple and inexpensive (Hameed et al., 2014).
The objective of this work was to study commercial ZnO and ZnO/ ZnFe 2 O 4 mixed oxides in the photocatalytic reduction of Hg 2+ present in synthetic effluent containing mercury chloride (HgCl 2 ) in solar and batch reactive systems.

Characterization of catalysts Thermogravimetric Analysis (TGA) and Differential Scanning Calorimetry (DSC)
The TGA and DSC analyses were performed with a Shimadzu TGA-50M Thermogravimetric Analyzer equipment and Metter Toledo Analytical Microbalance model MX5. A solid mass of 1g was used in the analyses. The temperature range used for the TGA analyses wes between room temperature and 1273 K. For the DSC analysis the temperature range was between room temperature and 773 K. The volumetric flow rate of the treatment gas, nitrogen (N 2 ), was maintained at 50 mL.min -1 , with a heating rate of 10 K.min -1 used in all experiments.

X-ray diffraction (XRD)
The diffractograms for calcined and non-calcined samples, prepared by the excess solvent impregnation method, as well as pure commercial oxide (ZnO) were obtained on a Bruker D8 Advance X-ray diffractometer using CuKα radiation (V = 40 KV, I = 35 mA). Data were analyzed according to JCPDS (Joint Committee on Powder Diffraction Standards) and ICDD (International Center for Diffraction Data).

Nitrogen Adsorption-Desorption Measurements (N 2 )
The specific surface area (Sg), pore volume (Vp) and average pore radius (Rp) as well as the adsorption isotherms, were determined on a Quanta Chrome model Nova 1200 using adsorption of N 2 at 77 K. The non-calcined and calcined samples had been previously submitted to a thermal treatment at 373 K and 573 K, for 8 hour under vacuum to eliminate any existing water within the pores of the solids.

Scanning Electron Microscopy (SEM)
For the micrographs a scanning electron microscope with a dispersive energy X-ray detector, Leo Electron Microscopy, model MEV Leo 440i, was used. For the X-ray spectra the acceleration voltage was 20 kV and the beam current at 700 pA. For coating the samples, a Sputter Coater Emitech metallic coating equipment, model K450, was used and the thickness of the gold layer was estimated at 200 A°.

Mössbauer spectroscopy
The analyses with Mössbauer Spectroscopy were performed at room temperature in a spectrometer (Halder Elektronik GmbH model MA 531). The equipment was calibrated with a metallic iron absorber, and measurements using 57 Fe as a nuclear probe were performed by means of 57 Fe γ radiation emitted by a 57 Co(Rh) source. For the Mössbauer spectroscopy measurements, the powder samples were encapsulated in small cylindrical containers (sample holder) of acrylic, with approximately 25 mg.cm -2 , an amount calculated to optimize this measurement.

Photoacoustic Spectroscopy
Photoacoustic spectroscopy was used to determine the minimum excitation energy of the conduction band electrons for the valence band, which correspond to the band gap energy. The analysis conditions were: 0.2 g of sample, frequency of 23Hz, sampling timing of 300 ms with 3 means and waiting time of 1s per measurement, slot of 3.16, BP/F/2F filter and gain of 12.

Reaction Systems
Calcined pure commercial oxide and catalysts, prepared by the impregnation method with excess solvent and calcined, were evaluated in the photocatalytic reduction of Hg(II). The reduction tests were carried out in the liquid phase in a bench scale photocatalytic reaction unit (using ultraviolet irradiation supplied by a 250 W mercury vapor lamp) and in a solar reactor. The reaction time was 3 hours to quantify the decrease in the concentration of Hg(II). During this reaction, samples were collected at regular intervals (0, 2, 5, 10, 15, 30, 60, 90, 120 and 180 minutes) and were filtered (0.45 μm Millipore membrane) for catalyst removal. The remaining concentration of Hg(II) in solution was determined by atomic absorption spectrophotometry (SpecrtAA -10 Plus VARIAN). 1 g.L -1 of catalyst was used for the photocatalytic tests.

Characterization of catalysts
The TGA / DSC profiles obtained for the mixed oxides containing 5 and 15% wt. Fe are shown Figure 1.
The TGA profiles indicated that the percentage mass loss is more evident for the sample containing the initial nominal metallic load of 15% wt. Fe, possibly due to the higher percentage of iron nitrate in the sample, thus resulting in a greater decomposition of the precursor. The mass loss observed from room temperature up to 400 °C is attributed to the H 2 O adsorbed on the catalyst surface. After 400 °C, the mass of the sample remains practically constant, indicating thermal stability from this temperature on. The results of DSC were also equivalent, and they detected endothermic and exothermic peaks, which justify the effects of the heat associated with chemical reactions and physical alterations, in which the endothermic peaks related to the loss of mass were associated with dehydration and decomposition, while the exothermic peaks explain the possible physical changes of the crystallization and oxidation reactions (Kanade et al., 2006).

B.
A.
The diffractograms obtained for samples containing noncalcined iron present a typical profile of amorphous material, which demonstrates the need for thermal treatment to arrange the atoms in the crystal and obtain a crystalline phase. The samples with 5 and 15wt% Fe in the ZnO matrix, submitted to the calcination process, presented the same crystallinity profile obtained for the pure ZnO. However, zinc ferrite peaks (ZnFe 2 O 4 ) were observed in the samples according to ICSD 22-1012 (Tadjarodi et al., 2015;Carvalho et al., 2017).
Similar results were obtained for the nominal catalysts metal loads of 8 and 10 wt% Fe. In fact, according to Qamar et al. (2017), the wet impregnation technique favors the interaction of Fe 3+ ions with the oxygen present on the ZnO surface, originating zinc ferrite.
Therefore, in Table 1, it is possible to analyze data of the interplanar space (d-spacing) and width at half height (FWHM). For the FWHM parameter, the results show that the variation obtained between the ZnO/ZnFe 2 O 4 and commercial oxide samples was approximately 0.01, while for the d-spacing parameter, a constancy of 2.47 was obtained. Thus, the difference obtained is too small to indicate any significant change in the ZnO crystalline structure due to the presence of zinc ferrite. According to Gonçalves et al. (2017), the differences observed may be due to defects in the structure of the semiconductor caused during the synthesis process, and these are responsible for small changes in the network parameters of the material under analysis. In addition, something similar is observed in the peak heights, indicating that the crystallite size samples are equivalent .   the ZnO crystals structure, which may be derived from the water used during the semiconductor synthesis (Raja et al., 2014). The vibrations between 3450 and 3500 cm -1 indicate the presence of hydroxyl OH groups (Maya-Treviño et al., 2015;Chakma et al., 2013;Khan and Khalid, 2010). According to Chakma et al., 2013 andPandiyarajan et al., 2012, the peaks that appear between 800 and 500 cm -1 can be attributed to Fe 3+ on the ZnFe 2 O 4 surface (Qamar et al., 2017). It is also observed that, with the iron percentage increase, the peaks corresponding to the H-O-H and Zn-O groups are minimized, while the band in the range 800-1500 cm -1 becomes more intense. Such behavior may be related to the fact that O-H clusters act as nucleation sites on which fine ferrite crystals can grow (Ameer et al., 2105). In the non-calcined sample different profiles of molecular vibration was observed comparing with the calcined samples.
In Figure 4B, the bands from 3700 to 3300 cm -1 are related to the symmetric and asymmetric stretches of the hydroxyl groups of H 2 O molecules and those located between 669 and 836 cm -1 correspond to the iron atoms  and at 1380 cm -1 , correspond to the stretching caused by NO 3- (Chakma et al., 2013;Khan et al., 2010;Maya-Treviño et al., 2015;Cunha et al., 1996).
The Mössbauer spectra and the hyperfine parameters for the ZnO/ZnFe 2 O 4 calcined at 400 o C, are shown in Fig 5 and Table 2, respectively.
The samples had well-defined double-dipoles with isomeric displacements of 0.34 mm.s -1 and quadrupolar displacement 0.47-0.49 mm.s -1 . which can be attributed to Fe 3+ ions dispersed in the octahedral structure. However, zinc oxide exhibits a hexagonal crystalline structure whereas the ZnFe 2 O 4 spinel presents two sites in the structure, one tetrahedral and the other octahedral (Tadjarodi et al., 2015;Chinnasamy et al., 2000).
Thus, by means of IS values and spectral area (A = 100%) it is assumed that all the iron present in the sample is contained in ZnFe 2 O 4 species (Zhu et al., 2016).
In Fig 6. it is possible to observe the N 2 isotherms obtained for the commercial ZnO and for the oxides modified with zinc ferrite. The ZnO and ZnO/ZnFe 2 O 4 10 sample profiles present the type II isotherm, indicating adsorption in superimposed multiple layers predominant in mesoporous structures according to the International Union of Pure and Applied Chemistry (IUPAC).
In the samples ZnO/ZnFe 2 O 4 5, ZnO/ZnFe 2 O 4 8 and ZnO/ZnFe 2 O 4 15, it was verified that the isotherms obtained are similar and predominantly of type V. The semiconductor isotherms also presented hystereses that are associated with mesopore forms. For ZnO/ ZnFe 2 O 4 5 and ZnO/ZnFe 2 O 4 samples, a tendency for N 2 hysteresis related to bottle-style mesopores is observed and the ZnO/ZnFe 2 O 4 15 sample shows conical from wedge associated with hysteresis type H3 (IUPAC).  It was observed that the volume of N 2 adsorbed by ZnO and ZnO/ZnFe 2 O 4 15 is higher than those obtained for the other samples, indicating that the interparticle porosity of these materials is higher in comparison to ZnO/ZnFe 2 O 4 5, ZnO/ZnFe 2 O 4 8 and ZnO/ZnFe 2 O 4 10.
Because the volume of N 2 adsorbed by the oxides can be correlated with the values of the specific area of the oxides, the greater the nitrogen adsorption, the greater the specific area of the material (Mandal et al., 2015).
The values obtained from the adsorption-desorption measurements of N 2 for specific surface area (Sg), pore volume (Vp) and average pore radius (Rp), as well as the results obtained from the photoacoustic spectroscopy for ZnO and ZnO / ZnFe 2 O 4 catalysts are shown in Table 3.
The photoacoustic spectroscopy result for the commercial ZnO was 3.18 eV. In the literature some authors (Mohamed et al., 2014;Ba-Abbad et al., 2013) found a band gap value of 3.37 eV for ZnO, while others (Aydin et al., 2013;Sakthievel et al., 2003) report band gap values for this semiconductor of 3.19 eV and 3.17 eV, respectively. In the ZnO/ZnFe 2 O 4 catalyst, a reduction of bandgap energy was generally observed. Based on the Mossbauer spectroscopy and X-ray diffraction results, the formation of heteroconjugated structures between Zn and ZnFe 2 O 4 is assumed, which justifies the extension of light absorption to the visible region (Wang et al., 2013).
Thus, the n-n heterogeneous structure characteristic of ZnO/ZnFe 2 O 4 is capable of efficiently separating the photogenerated electrons from the gap, resulting in imporved photoactivation under sunlight (Rahimi et al., 2015). This is because the electrons in the valence band (VB) of ZnFe 2 O 4 are preferentially excited to their conduction band (CB) under visible light irradiation and therefore generate an equal amount of holes in their VB. The CB level of ZnFe 2 O 4 is lower than that of ZnO (0.9 eV). Thus, the photogenerated electrons in the CB of ZnFe 2 O 4 can be transferred to that of ZnO. On the other hand, the photogenerated holes are transfered from the VB of the ZnFe 2 O 4 to the VB of the ZnO, because the VB level of the ZnO is 2.5 eV higher than the VB of ZnFe 2 O 4 . Therefore, the probability of electron-hole recombination can be reduced.

Adsorption and photocatalysis tests of Hg(II) reduction
Batch reactor tests were performed in the absence of UV light, before the photocatalytic reduction reactions, to verify the adsorption of mercury(II) on the commercial oxide (ZnO) surface, as well as on   Fig. 7. The pure ZnO, when present in a medium containing an aqueous solution and 120 mg.L -1 of mercuric chloride, did not show the ability to adsorb Hg(II), and the same was observed for an aqueous solution containing mercuric chloride and oxygen (0.83% removal of Hg from the medium). In the presence of the Hg(II) aqueous solution containing formic acid and oxygen, the adsorptive capacity of the semiconductor in 3 hours was 22% and 19%, respectively.
For the ZnO/ZnFe 2 O 4 5 catalyst, the adsorption percentages of 44.45%. 31.24%. 22% and 19.3% were observed for conditions C1. C2. C3 and C4, respectively. This profile is maintained for the other catalysts. The highest removal percentage was verified for the oxygen-containing solution (33.62%) for calcined ZnO/ZnFe 2 O 4 8 and was also observed for E.
calcined ZnO/ZnFe 2 O 4 15 (41%). For calcined ZnO/ ZnFe 2 O 4 10, the highest adsorption capacity was obtained for the C4 reaction condition, with a removal of approximately 32%. The pH is one of the factors that most affects the adsorption capacity of metallic ions on the surface of ZnO, because it affects not only the adsorbent structure, but also the adsorbate (Sheela et al., 2012). In addition, heavy metal adsorption is attributed to different ion exchange mechanisms. The metal ions can move through the catalyst pores, the channels present in the crystalline structure, and may also substitute hydroxyl groups on the semiconductor surface.
In relation to mixed oxides, it was reported that, in aqueous media, the surface of ferritic nanoparticles in the spinel form has a positively or negatively charged surface for acidic or alkaline conditions, respectively (Kefeni. et al., 2017).
The photocatalytic results using pure ZnO and the ZnO-based catalysts prepared by the excess solvent impregnation method are shown in Fig 8. According to the results obtained, the catalyst that presented the best performance was pure ZnO, when the reaction was carried out in the reaction medium containing formic acid as hole scavenger. Under these conditions, the mercury photoreduction was fast, with a 100% reduction in the detection limits of the flame atomic absorption spectrometry (FAAS). It was not possible to detect the presence of Hg(II) in the reaction medium after 10 minutes of reaction. In the presence of the aqueous solution containing only mercuric chloride at the concentration of 120 mg.L -1 (C1) and the aqueous solution containing HgCl 2 and oxygen (C3), a complete reduction of the Hg(II) was observed in 30 minutes of reaction, while in the presence of the aqueous solution containing mercuric chloride, oxygen and 10 mM formic acid (C4), the time required for the complete reduction of Hg (II) was 15 minutes.
The photocatalytic activity obtained for ZnO can be attributed to the high number of defects present in its crystalline structure -oxygen, interstitial zinc and interstitial oxygen vacancies -which permits the incorporation of impurities, contaminants or dopants in its reticulum (Kayaci et al., 2014;Ansari et al., 2013). Ong et al. (2018) affirmed that the oxidation/ reduction reactions by photocatalysis are dependent on the bandgap energy as well as the OH radicals present on the catalyst surface.
The best performance of the commercial oxide regarding the mixed oxides can be justified based on the results of FTIR. For the ZnO/ZnFe 2 O 4 materials, a decrease of the bands corresponding to the H-O-H group compared to ZnO is observed. Such groups may correspond to the adsorption sites of Hg(II), limiting one of the steps of the photocatalystic mechanism.
For the ZnO-based catalysts containing different nominal iron metal charges (ZnO/ZnFe 2 O 4 ), there was a complete removal of Hg(II), although the average time required for this was between 30 and 90 minutes of reaction. The ZnO/ZnFe 2 O 4 5 and ZnO/ZnFe 2 O 4 catalysts showed a better performance when the reaction medium was composed of an aqueous solution containing mercury, formic acid and oxygen (C4), in which the time required for complete reduction of Hg(II) was 30 minutes. For the catalysts ZnO/ZnFe 2 O 4 10 and ZnO/ZnFe 2 O 4 15, the reaction medium containing mercuric chloride and formic acid (C2) was the most appropriate one, the time required for the photoreduction of the mercury ion being 30 and 60 minutes, respectively.
Among the synthesized materials, the best performance of the ZnO/ZnFe 2 O 4 5 sample can be justified on the basis of the adsorption and desorption data of N 2 . Despite not having the largest specific surface area of the mixed oxides, this sample presents the largest average pore radius. This characteristic may allow better transport from Hg(II) adsorption sites favoring the photocatalytic reduction (Li et al., 2014). According to Cheng et al. (2017), ZnO/ZnFe 2 O 4 semiconductor materials that present a high specific area provide a greater amount of active sites for the photocatalytic process, but this characteristic generates a greater amount of defects in the structure of the catalyst, which facilitates the recombination of the electrons and the holes, inhibiting the physical-chemical transformations in the substrate .
In addition, it is possible that the average pore volume of magnitude 0.0032 cm 3 .g -1 of the ZnO/ ZnFe 2 O 4 sample favors better light transmission to the inner cavities of the catalyst (Li et al., 2014). Regarding the presence of sacrificial agents, it is observed that they had a limited contribution in the process of reduction of mercury (Hg 2+ ) present in the initial solution. According to studies (Aguado et al., 1995), oxygen exerts the role of inhibitor in the photoreduction process of Hg(II) as it competes with the electrons photogenerated by the semiconductor material. However, formic acid has a counterpart function, reacting with the generated holes, favoring the reduction of mercury (Cai et al., 2016;Wang et al., 2004).
The ZnO and ZnO/ZnFe 2 O 4 5 oxides were used for the catalytic photoreaction experiments under solar radiation. Thus, it was observed that there was very little mercuric ion reduction (Fig 9), which may be linked to E.
the bandgap of 1.94 eV -and can be attributed to the presence of ZnO 2 Fe 4 -providing a rapid recombination of the electron/hole pairs (e -/h + ) and reducing the quantum efficiency of the material. This problem can be solved by doping the zinc ferrite with Ag, as related in the literature (Harish et al., 2012), or with other materials such as Ca and Gd (Bini et al., 2018). The photocatalytic performance is significantly influenced by the reaction conditions, such as: incident light spectrum and intensity, reaction temperature, amount of catalyst, the addition of cocatalysts, the pH of the solution, reactor design and configuration, and so on. Therefore, it is a difficult task to compare the efficiencies of photocatalysts reported under different conditions (Xiao et al., 2018).
In this context, Table 4 indicates reports of different experimental parameters and their results.

CONCLUSIONS
The pure oxide ZnO presented Hg(II) photoreduction efficiency in the all experimental conditions and reaction systems used in this work. It was also observed that, for reactions carried out in batch reactor in the presence of formic acid, the complete reduction of Hg 2+ was achieved more quickly. The results indicated that all the ZnO/ZnO 2 Fe 4 catalysts were able to reduce Hg(II) under UV light, but a lower efficiency was observed in the photoreduction process of Hg(II) when compared to pure oxide.  Table 4. Best results of several studies which report Hg(II) photoreduction in an aqueous phase. *Addition of hole scavenger; ** absence of hole scavenger.