Azo dyes are present in wastewaters from the textile, paper, plastics, leather, food, cosmetics and some other industries. Unfortunately, they constitute one of the major sources of environmental degradation of water (Pavithra et al., 2019). Azo dyes are not only dangerous to aquatic life, but in many cases carcinogenic to humans and animals (Fatima et al., 2017). Researchers are thinking about creating an effective technique for the removal of pestilent dyes from wastewaters. Several methods like adsorption (Ayati et al., 2016), hydrogenation (Fu et al., 2019), photocatalysis (Yang et al., 2018, 2019; Chen et al., 2019; Zeng et al., 2019; Feng et al., 2020), degradation  by  biological  (Rajasulochana  and  Preethy, 2016) and chemical oxidation (Collivignarelli et al., 2019; Katheresan et al., 2018) have been applied for industrial wastewaters remediation before their disposal to the environment. The use of hydrogen peroxide as an oxidizing agent was also proposed for a long time but proved to be ineffective when the load of dye is high (Wang et al., 2016). So, research groups have been focused on advanced oxidation processes, some environmentally friendly methods allow treatment of dyes in wastewater. These technologies have already shown their potential in treating toxic and biologically recalcitrant organic pollutants. The advanced oxidation processes include O₃ (Biard  et  al.,  2018),  UV  (Seo  et  al.,  2019), UV/H₂O₂, (Li et al., 2020), O₃/UV (Wardenier et al., 2019), and Fenton (Bello et al., 2019; Zhu et al., 2019; Xiao et al., 2020). Among all advanced oxidation processes, Fenton reaction using soluble iron salts in Fe2⁺ or Fe³⁺ form and H₂O₂ can produce powerfully oxidizing radical species such HO^• (Fe ²⁺ + H₂O₂ → Fe3⁺ + HO^• + OH⁻ (1)  Fe³⁺ + H₂O₂ → Fe²⁺ + HO₂^• + H⁺ (2)) for decomposition of dyes. However, the implementation of homogeneous Fenton oxidation based on ferrous or ferric salts has several limitations, including the conversion of Fe(OH)₃ into sludge at a pH above 4 (Zhu et al., 2019) which requires an additional treatment (Cetinkaya et al., 2018) and a cost for their elimination. To improve the oxidation efficiency heterogeneous Fenton-like reaction has been developed. Heterogeneous Fenton-like processes can be established by replacing Fe²⁺ in the Fenton reagent with a solid catalyst including metal oxide. According to literature (Thornburg et al., 2016; Ziolek et al., 2015; Ziolek and Sobezak, 2017) reactive oxygen species can be generated in situ on the surface of selected metal oxides (such as Nb₂O₅, V₂O₅, TiO₂, Fe₂O₃) by the treatment with hydrogen peroxide, and then successfully used as heterogeneous catalysts for degradation of organic pollutants (Bello et al., 2019; Wiedmer et al., 2016; Zhu et al., 2019). The advantages of using metal oxide as solid catalyst in heterogeneous Fenton-like process comparatively to homogeneous Fenton process are: (a) The reactions shows performance in a wide range of pH (Zhu et al., 2019) (b) metal oxides can be recovered easily; (c) metal oxides decompose hydrogen peroxide H₂O₂ to hydroxyl radical HO^• like in classical Fenton reaction, and prevent precipitating of iron hydroxide (sludge) which reduce the efficiency of global oxidation process (Zhu et al., 2019). Among the heterogeneous Fenton-like processes, Photo-Fenton has been developed as a promising method using catalysts such Fe₂O₃ (Domacena et al., 2020), Fe₃O₄, (Arshad et al., 2017) FeOOH (Liu et al., 2017), Fe₂O₃/Al₂O₃ (Di Lucia et al., 2018) to overcome the classic homogeneous Fenton drawbacks mentioned above. However, UV and Visible light-based photo Fenton oxidation also has practical limitations. At neutral pH, a photo Fenton process showed low efficiency and needed the addition of iron complexing agents (Ahile et al., 2019). Consequently, significant efforts to develop various new heterogeneous Fenton catalysts have been made including; CuO (Deka et al., 2016), CeO₃ (Wei et al., 2019) and Mn₃O₄ (Osgouei et al., 2020). The aim of this work is to propose a new alternative metal oxide such as tungsten oxide coupled with H₂O₂ for the treatment of methyl orange. WO₃ can be a remarkable candidate as solid Fenton catalyst because in nano size, transition metal oxides has a great interest for oxidation due to is large surface area, which exhibits many advantages in the chemical catalysis field, such as a low diffusion resistance, easy accessibility to reactants, and large number of active sites (Gadipelly and Mannepalli, 2019).  The  use  of  WOx/H₂O₂  has  not been reported in the literature as heterogeneous Fenton-like process. Thus, to evaluate the outstanding efficiency using the WO₃ catalyst, a comparative study with other metal oxides for removal methyl orange from aqueous solution was presented.

Experimental

Tungsten oxide NPs were prepared by a simple dispersion of metallic tungsten powder (1.0 g) (APS 1-5 μm, purity 99,9%, Alfa Aesar Co.) in deionized water (10 ml), glacial acetic acid (1 ml , ACS reagent, purity ≥99.7%) followed by addition of H₂O₂ (10 ml, aqueous solution; 30%, Fisher Scientific Co.) dropwise. The mixture was kept at 0°C in an ice bath. The oxidative dissolution process was very exothermic; therefore it was realized carefully under a well-ventilated fume hood. The solution was put in the glass reactor equipped with a condenser and heated at 85 °C for 3h. A precipitate is formed and centrifuged at 8000 rpm. Hydrated tungsten oxide WO₃-H₂O material was obtained without any surfactant by washing the precipitate with water and ethano,l then dried at 60°C overnight (16 h). Hydrated tungsten oxide WO₃-H₂O material followed two treatments: it was annealed at 200 and 600°C for 4 h and obtained h-WO₃ and WO₃ respectively.

Characterization

Powder X-ray diffraction (XRD) patterns of the samples were obtained on a Bruker D8 Advance X-ray diffractometer equipped with a Cu Kα radiation source (λ = 1.5418 Å). Fourier transform infrared (FTIR) absorption spectra were measured with an FTS 45 infrared spectrophotometer using KBr pellet technique. Scanning electron microscopy (SEM) images were taken on a JEOL 6360 instrument at an accelerating voltage of 3 kV. X-ray photoelectron spectroscopy (XPS) measurements were performed on a VG Escalab 220i-XL using a polychromatic Al source (hν = 1486.6 eV) operating at 15 kV and 20 mA. The signal was filtered with a hemispherical analyser (pass energy = 20 eV) and the detection was performed with a multi-channel detector. The base pressure inside the spectrometer during analysis was 3 × 10^-10 Torr. 

Catalytic tests

Catalytic tests were carried out in a reactor capacity of 250 ml attached with a condenser. 100 ml of aqueous solution of methyl orange was introduced and 50 mg of various tungsten oxides obtained following 2.5 ml of H₂O₂. The reactor was put in the oil bath previously heated and maintained at 65°C. During the reaction, the mixture was continuously stirred using magnetic stirrer (agitation rate ~ 400 rpm) for 3 h. The residual concentration of dye in the aqueous solution was determined each 20 min via UV-vis spectrophotometer at 464 nm. At the end of the reaction time, the tungsten metal oxide particles were separated from aqueous solution by using centrifuge at 8000 rpm during 10 min. Tungsten metal oxide particles were recovered then washed with ethanol and water several times and dried at 60°C overnight for reusability.

XRD  analysis  is   considered   a   promising   technique, investigating the crystalline structure and size of the synthesized nanoparticles. Figure 1 displays the XRD pattern of the synthesized tungsten oxide nanoparticles before and after calcination.  The precipitate obtained showed various diffraction peaks at 2θ = 16, 53; 25, 63; 35, 03 and 49, 64 which are assigned to (020), (111), (131) and (202) reflections according to the JCDPS card N° (84-0886). This crystalline structure is identical to hydrated tungsten oxide WO₃-H₂O of orthorhombic structure. This type of structure was also observed in the case of hydrothermal and solvothermal synthesis of tungsten oxide nanoparticles for sensing volatile organic compound (Perfecto et al., 2016) and photocatalytic activity (Liu et al., 2019), respectively. The calcination of WO₃-H₂O at 200 and 600°C respectively showed different XRD patterns. At 200°C diffraction peaks were in agreement with the hexagonal phase of WO₃ (JCDPD card N° 75-2187). The h-WO₃ hexagonal phase was also observed after calcination treatment (Perfecto et al., 2016). The peaks corresponding to (100), (001), (200) and (2011) crystal planes were also identified. Increasing the calcination temperature up to 600°C showed the peaks assigned to WO₃ monoclinic structure (JCDPD card N₀71-2141). These peaks appeared at 2θ values of = 23.12; 23.58; 24.38; 26.59; 28.93; 33.26; 34.17; 41.9 having miller indices (002), (020), (200), (123) (112), (022), (220), (222) respectively (JCDPD card N₀71-2141). This monoclinic crystalline structure had been shown at this  temperature   previously   (Kang   et   al.,  2020)  and increasing the calcination temperature up to 450°C produced a stable monoclinic phase (Salkar et al., 2018).

In order to investigate the surface morphology of these tungsten oxide particles, scanning electron microscopy was carried out for the various samples. Figure 2 shows morphology of the tungsten oxides particles.

Figure 2A shows the nanoplates obtained before calcination. The shape evolved with the temperature of calcination, giving nanorod-like structure (Figure 2B) at 200°C and pallets with large thickness at 600°C (Figure 2C). The particles are not uniform in micro or nano size but regular without addition of surfactant. The X-ray photoelectron spectroscopy (XPS) is a useful tool to investigate the composition and oxidation states of elements presented over the surface of the tungsten oxide particles. Figure 3A shows a wide survey scan of the samples, clearly indicating the presence of W and O elements. The deconvolution of W4f spectrum in Figure 3B clearly shows two peaks at 35.6 and 37. 7 eV which were assigned to W 4f_7/2 and W4f_5/2 respectively for WO₃-H₂O. The binding energy at 35.7 and 37.8 eV were assigned to W4f_7/2 and W 4f_5/2 for h-WO₃, while the peaks located at 35.4 and 37.5 eV were attributed to W 4f_7/2 and W 4f_5/2 of monoclinic WO₃.

XPS analysis is one of the fundamental tool and the Figure 3B depicted, two spin-orbit doublets corresponding to the different valence states of W. This spin-orbit splitting of W4f is 2.1 eV for all tungsten oxides, which  agrees  with previously published literature for W⁶⁺ ions or +6 oxidation state (Kolhe et al., 2020). Figure 3C shows O1s spectra. The deconvolution of these peaks showed various weak peaks at different binding energy only for h-WO₃ and WO₃-H₂O. The monoclinic tungsten oxide structure showed a peak at 530 eV and both h-WO₃ and WO₃-H₂O samples at 530,2 eV which are attributed to the lattice oxygen bonded with W (Parthibavarman et al., 2018). WO₃-H₂O showed a high peak of adsorbed H₂O at 532.5 eV which confirms XDR results. The peak of adsorbed H₂O shifted to a weak at 532.4 eV due to a very small amount of water containing h-WO₃. The peak at 531.4 eV is due to OH groups. XPS study strongly supported the formation of WO₃ and confirmed XRD patterns observed for tungsten oxide structures. The same results were observed by several authors during the synthesis of tungsten oxide (Thornburg et al., 2016; Kolhe et al., 2020; Parthibavarman et al., 2018). The small characteristic peak of the carbon element is observed at 284 eV in Figure 3A and is attributed to the carbon dioxide of the air or to the impurities of the reagents or to the washing solvent. These same results were   demonstrated   by   Geng   et   al. (2017). Figure  4 shows the FT-IR spectra of the various tungsten oxide synthesized. 

Identification of molecular structures reveals normal vibration modes by stretching (ν), plane bending (δ). The vibration bands are mainly the fundamental vibrations of the W=O, W-O and W-O-W chromophores and the stretching of O-W-O bonds (Ziolek et al., 2015). Below 1000 cm^-1, the FT-IR spectra showed significantly various vibration peaks at 942 and 671 cm^-1 assigned as the stretching of W=O and O-W bonds in the hydrated tungsten oxide WO₃-H₂O crystalline structure. These stretching bonds exist also in both hexagonal and monoclinic structure. The increase of the calcination temperature up to 600°C created a characteristic peak of O-W-O stretching vibrations in the WO₃ monoclinic crystal lattice.

The absence of carbonyl peak around 1730 cm^-1 indicated that no carbon contained all the samples, which confirmed the impurities shown in XPS spectra (Figure 3A). For hydrated tungsten oxide, the symmetric stretching vibrations related to WO₃ were observed and the  intercalated  water  molecules   (W-OH ... H₂O)   are characterized  by  a  significant  peak  at  3397  cm^-1.  The bending  vibration  is  observable  at   approximately  δ  = 1613 cm^-1 which corresponds to the W-OH plane for the WO₃-H₂O. These vibrations generate the presence of H₂O, which confirms the orthorhombic WO₃-H₂O structure observed in XRD and H₂O absorbed in XPS. Indeed, WO₃-H₂O was obtained at 85°C without calcination, that did not allow removal of water molecules; thus creating their inter reaction with tungsten oxide structure. The peak at 3397 cm^-1 is weak and not significant for the h-WO₃ comparatively to WO₃-H₂O structure. The bending vibration is observable at approximately δ = 1613 cm^-1. The h-WO₃ shows that water molecules are removed but there is a small amount intercalated water molecules containing on the plane. The water intercalated was shown by XPS (Figure 3C). Increasing the calcination temperature up to 600°C removed water and led to obtain non-hydrated crystal structure WO₃ because there are no vibration peaks in the range of 3000-3700 cm^-1 corresponding to the presence of water. These results confirmed structures proposed by XRD and XPS.

Catalytic activity

The catalytic activities of tungsten oxides nanoparticles synthesized were carried out on the capacity of methyl orange remove from an aqueous solution. Efficiency of the removal process was monitored by UV-visible spectroscopy. Figure 5 shows the methyl orange removal rate, using tungsten oxide nanoparticles with hydrogen peroxide.

Hydrated tungsten oxide WO₃-H₂O alone eliminated around 3% of methyl orange after 3 h. The amount of methyl orange removed, increased up to 10% when H₂O₂ was also used only, slightly high comparatively to hydrated tungsten oxide WO₃-H₂O. According to the literature, it is possible to oxidize many organic compounds with hydrogen peroxide (Zhang et al., 2019). Coupling H₂O₂ and the tungsten oxides nano particles increased the removal efficiency of methyl orange. The values reached 20, 30 and 47% for the coupling of hydrogen peroxide with WO₃, h-WO₃ and WO₃-H₂O catalysts respectively. The use of hydrogen peroxide with tungsten oxide nanoparticles had a benefit effect for methyl orange removal. Many authors had shown that the catalytic activity of metal oxides in dye degradation is enhanced when metal oxides are combined with hydrogen peroxide in acid pH medium (Ziolek and Sobezak, 2017; Osgouei et al., 2020; Deka et al., 2016). The reaction of these tungsten oxide nanoparticles with H₂O₂ led to generation of strongly reactive oxidative species such as O₂^•- and O₂^2- , HO^• (Ziolek et al., 2015). All the catalytic tests were carried out at pH = 6.9 that is very difficult in classical Fenton process. The coupling of WO₃-H₂O with hydrogen peroxide displayed a high performance comparatively to h-WO₃ and WO₃ respectively. Therefore, WO₃-H₂O led to formation of more reactive oxygen species than h-WO₃ and WO₃. The presence of water in crystal lattice increased the ability of WO₃-H₂O to generate more oxidative species. Thus, the high  production  of  hydroxyl  radicals  in the presence of water by the hydrated orthorhombic structure of tungsten oxide caused significant degradation of methyl orange comparatively to the hexagonal and monoclinic structures respectively. The reaction of H₂O₂ with the metal oxides is very complex, but the mechanism for production of hydroxyl radicals is known and consists of two steps: adsorption of H₂O₂ on metal oxide followed by its decomposition or cleavage to form hydroxyl radicals (Lousada et al., 2015). These two steps are also governed by the state of metal oxide surface and the kinetics of decomposition respectively (Lousada et al., 2015).

The hydrated structures of tungsten oxide promoted the kinetics of hydrogen peroxide cleavage for the production of hydroxyl radicals. It is known that the kinetics of decomposition is strongly influenced by the pH of the reaction medium (Lousada et al., 2015; Wolski et al., 2019). To increase the performance of hydrated tungsten oxide, the catalytic reaction tests were done at a pH range as shown in Figure 6. The methyl orange removal rate increased when the pH medium decreased and maximum of methyl orange removal reached up 51% when the pH medium is 4. The results confirmed that shown in precious literature (Lousada et al., 2015; Wolski et al., 2019). The mass effect was also realized. Figure 7 displayed that the high rate of methyl orange removal was observed when the mass is 300 mg for WO₃-H₂O and WO₃. At this mass value, the methyl orange  removal reached up to 89.7 and 74.4% for WO₃-H₂O and WO₃ respectively. Thus, increasing the mass of tungsten oxide led to considerable removal of methyl orange from water. The UV-vis absorption spectra of methyl orange were recorded and shown in Figure 8. The spectrum of initial methyl orange shows two peaks located at 464 nm from the azo bond and at 298 nm attributed to benzene nucleus in the methyl orange molecule before treatment. After the end of reaction time of degradation, the decrease in these peaks was observed when H₂O₂ is used, indicating that the azo bond was destroyed. The treatment of methyl orange with the coupling WO₃-H₂O/H₂O₂ showed that the absorption peak from the azo bond decreased considerably, while the absorption at 298 nm from the benzene nucleus increase drastically and shifted to 288 nm, indicating that degradation intermediates containing the benzene ring were generated. These results are consistent with the work reported in previous literature (Chen et al., 2015; Omri et al., 2020).

The efficiently of monoclinic WO₃ is also increased. It was essential to compare the efficiency of the heterogeneous Fenton-like system using tungsten oxide for the removal of methyl orange with another heterogeneous Fenton-like system using other metal oxides. The Table 1 showed the performance of the Fenton-like process obtained with different metal oxides for  the  removal  of  methyl  orange  only.  As  showed  in Table 1, the results of the study are higher than iron oxide, titanium oxide nanoparticles capacities. The supported metal oxides showed  higher  performance.  All of these others metal oxide used photo Fenton as a Fenton-like process which used UV or Visible light to increase   radical    hydroxyl    amount.   The   Fenton-like process using tungsten metal oxide nanoparticle requires no UV input and showed high performance at pH= 4 in comparison to these other metal oxides, which need more strong acid pH medium.

Reusability of the tungsten oxide nanoparticles was investigated. Unfortunately, it was difficult to do more than 2 cycles of methyl orange removal with WO₃-H₂O because of the loss of hydrated tungsten oxide particles as well as the leaching. As shown in SEM analysis, WO₃-H₂O  showed  the  small  size  particles  comparatively  to monoclinic WO_3. It was then, not so easy to recover these particles. However, monoclinic WO₃ particles were recovered easily and reused in 4 consecutive runs. As shown in Figure 9, the removal efficiencies of WO₃ found are 74.4; 73. 2; 71.3 and 69.6% in first cycle, second, third and fourth cycle respectively. A slight loss was observed in efficiency of monoclinic tungsten metal oxide nanoparticles. Monoclinic WO₃ showed excellent stability even after four continuous cycles of activity, whereas the removal    efficiency   of    monoclinic   WO₃   is   relatively reduced after four cycles of reusability.

This study demonstrated that the combination of tungsten oxide nanoparticles with hydrogen peroxide as a Fenton-like system improved the removal of methyl orange from water. This process did not need pH = 3 condition to show high performances. The tungsten oxide particles in nano size have been well identified and characterized by various technical. Coupling the tungsten oxide nanoparticles with H₂O₂ showed high methyl orange removal rate comparatively to hydrogen peroxide or tungsten oxide used alone. At pH = 4, the increase of the mass of tungsten oxides nanoparticles produced better conditions with H₂O₂ to form more reactive oxygen species attributed to orthorhombic tungsten oxide WO₃-H₂O comparatively to h-WO₃ and WO₃. The reusability of the monoclinic WO₃ was done and showed good stability. The Fenton-like process using WO₃/H₂O₂ showed higher capacity to remove methyl orange than many other metal oxides associated with hydrogen peroxide.

The authors have not declared any conflict of interests.

The authors appreciate the government of Côte d’Ivoire under the Research Program of Merit Student for the support. The authors also acknowledge the Institut National de Recherche Scientifique-Énergie, Matériaux et Télécommunications (INRS-EMT) of Canada and the Institut de Science des Matériaux de Mulhouse (IS2M), of France for their cooperation.