Nb2O5 monolith as an efficient and reusable catalyst for textile wastewater treatment

In this work, the degradation of methylene blue (MB) in aqueous solution by a heterogeneous Fenton-like process was studied. Nb2O5 supported on an anodized aluminum monolith was used as catalyst. The Nb2O5 coating generated during the catalyst synthesis was uniform and porous, with a thickness of 7 μm. Response surface methodology was used to study the relationship between process variables and MB degradation and chemical oxygen demand (COD) reduction. The optimal operating conditions were: initial MB concentration of 45 mg L , initial H2O2 concentration in the reaction medium of 40 mM, catalyst loading of 7.85 kgNb2O5 m − , and pH 4, reaching a MB degradation of 98% and a COD reduction of 79% at 150 min. The high activity of the Nb2O5 monolith catalyst remained constant after 10 catalytic tests.


Introduction
Textile industries are responsible for serious environmental problems due to the large volume of dyestuff wastewater that they generate (200 to 350 m 3 of wastewater per ton of finished product) [1]. Textile wastewaters commonly present high values of pH, biochemical oxygen demand (BOD), chemical oxygen demand (COD), turbidity, and occasionally potential carcinogenic or mutagenic substances [2]. Colored wastewater also reduces the light penetration through the water surface, and consequently the photosynthetic activity of aquatic organisms [3].
Nowadays, conventional technologies applied for textile wastewater treatment are based on coagulationflocculation processes, activated carbon adsorption and membrane filtration technique. These methods are nondestructive, transferring the organic contaminant to another medium that demands further treatment, thus increasing the processing costs. On the other hand, biological wastewater treatment methods generally have a low efficiency in color removal due to the high resistance of the organic dye compounds to degradation [2].
In this context, advanced oxidation processes (AOP) using heterogeneous catalysts are considered a promising technology for the discoloration of textile wastewater, as they lead to the total mineralization of most organic pollutants without generating solid wastes [4]. In recent years, there has been a growing interest in the application of niobia (Nb 2 O 5 ) for water remediation due to its acid sites properties, good chemical stability, null toxicity and commercial availability [5].
The application of niobium pentoxide nanoparticles for dye photodegradation has been previously studied, exhibiting promising performances [6,7]. Kumari et al. [8] reported a dyestuff degradation (methylene blue, MB) of between 40 and 70% using Nb 2 O 5 nanoparticles as photocatalyst. de Carvalho et al. [6] observed almost 100% MB degradation using niobium pentoxide as photocatalyst after 3 h of sunlight exposure. In contrast, dyes in high concentration limit the light penetration into the solution, increasing the light scattering effect, and decreasing the process efficiency [9].
Niobium pentoxide is also a suitable catalyst for oxidation reactions using hydrogen peroxide (H 2 O 2 ) as oxidant [10]. Ziolek et al. [11] reported that Nb 2 O 5 contacted with hydrogen peroxide exhibited high ability to form radical peroxo species, generating active oxygen for the alcohol oxidation. On the other hand, textile wastewater treatments using Nb 2 O 5 catalyst, in nonphotocatalytic systems, have been scarcely reported [12]. Oliveira et al. [13] observed that the presence of hydrogen peroxide enhances the generation of hydroxyl radicals, and consequently, the dyestuff oxidation. Wolski et al. [14] reported a 51% degradation of Rhodamine B in 30 min using Nb 2 O 5 with hydrogen peroxide, observing similar results, in terms of degradation efficiency, with the photocatalytic system at lower dye concentration. This appreciation is in good agreement with the findings documented by Ucker et al. [15].
No reports could be found in the literature on the effects of the operating conditions for textile wastewater treatment using a Nb 2 O 5 monolithic catalyst in the presence of hydrogen peroxide. The use of structured catalysts is affordable considering that they can be reused in multiple cycles, overcoming the difficulty of separating the powder catalyst from the reaction medium, thus reducing processing costs.
In this paper, the oxidation of MB (as model dye compound) over an anodized aluminum-supported Nb 2 O 5 catalyst in a monolithic stirrer reactor was studied. The effects of key parameters (catalyst loading, initial dye concentration, pH and initial hydrogen peroxide concentration) on the dye degradation and the stability of the monolith catalyst were evaluated.

Catalyst synthesis
For the preparation of the anodized aluminum monoliths, an Al1050 commercial aluminum sheet was used as substrate (the composition of the aluminum sheets is given in Table S1 in the Supplementary Information). The aluminum sheets were washed with distilled water and acetone, then the anodizing process was carried out using a 1.6 M H 2 C 2 O 4 (≥ 99% purity; Sigma-Aldrich) aqueous solution as electrolyte and a power density of 2 A dm − 2 at 40°C, with 40 min of anodizing and 40 min of pore opening. After anodizing, the samples were washed with distilled water, dried and calcined at 450°C for 120 min using chromatographic grade air (99.999% purity) with a flow rate of 20 mL min − 1 . The monoliths were prepared by rolling around spindle alternate anodized a smooth and a corrugated sheet conforming sinusoidal channels. The final monolith was a cylinder of 14 mm diameter and 15 mm height (the geometric dimensions of the monolith are shown in the Table S2 in the Supplementary Information). Nb 2 O 5 (99% purity), provided by Brazilian Company of Minerals and Mineralization (CBMM), was dissolved in 1 M oxalic acid aqueous solution. The Nb 2 O 5 monolith catalysts were synthesized by wet impregnation using a niobium aqueous solution (100 mM, pH 4) for 60 min. After impregnation, the samples were dried at room temperature for 48 h and finally calcined at 550°C for 3 h in air flow (AGA chromatographic grade). Three impregnation, drying and calcination cycles were performed. The monolith catalysts are referred to as Nb 2 O 5 /Al 2 O 3 /Al.

Catalyst characterization
The Nb 2 O 5 /Al 2 O 3 /Al catalysts were characterized using N 2 adsorption-desorption isotherms at − 196°C for surface area measurements. The specific surface area of the catalysts was calculated according to BET method. The pore volume and pore size distribution of the samples were calculated from the N 2 adsorption isotherm with the BJH method, using a Quantachrome Instruments NOVA 1200e analyzer. Before the analysis, each sample was degassed in vacuum at 120°C for 24 h.
The Nb 2 O 5 loading in the monolith catalysts was determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) with a Shimadzu 9000 spectrometer. Before the measurement, the samples were dissolved in aqua regia.
The catalyst morphology was characterized by scanning electron microscopy (SEM) on a LEO-EVO 40 XVP microscope. The samples were previously coated through a metallization process.
The adherence of the Nb 2 O 5 layer deposited over the anodized aluminum surface was tested by the ultrasonic method. The monoliths were immersed in 50 mL of distilled water and subjected to an ultrasound bath at room temperature for 120 min in a Cole-Palmer 8892E-MT (47 kHz, 105 W) apparatus. Then the monoliths were dried. The weight of each sample was measured before and after the ultrasound test to determine the Nb 2 O 5 layer adherence, which was calculated as the percentage ratio of the amount of coating material retained on the substrate.
X-Ray diffraction (XRD) was applied to identify the crystalline phases of the catalyst. XRD patterns were obtained using a Rigaku D-max III-C diffractometer, with a monochromatic Cu Kα (l = 0.154060 nm) operating at 35 kW and 30 mA.

Reaction test
The reaction tests were carried out at 20 ± 1°C in a Pyrex batch reactor equipped with a monolith stirrer, in the dark. A schematic diagram of the monolithic stirrer reactor used in the reaction tests and its characteristic parameters are given in Fig. S1 and Table S3 in the Supplementary Information. The duration of the reaction test was 150 min with an agitation rate of 500 rpm. In each test, 250 mL of MB (≥ 97% purity; Sigma-Aldrich) solution were used, and the catalytic monoliths were mounted on the axis of the stirrer. The pH of the reaction medium was regulated by adding 0.1 M HCl or 0.1 M NaOH solution (≥ 98% purity, Sigma-Aldrich reagents). The reaction progress was monitored using a UV-VIS spectrophotometer (PG Instruments, T60). The wavelength selected for the measurement was 663 nm (maximum absorbance for MB), and the MB concentration was calculated using a calibration curve [9]. The MB degradation during the reaction test was determined using Eq. (1): where ABS 0 is the initial absorbance of MB solution, and ABS F is the final absorbance (after the reaction test). The efficiency of the oxidation process was also evaluated in terms of COD reduction as follows: where COD 0 is the initial COD (untreated sample) and COD F is the COD of the sample treated by the advanced oxidation process. COD analysis was carried out using commercial HACH vials (Method 8000). The samples were prepared using sulfuric acid and potassium dichromate as oxidizing agent at 150°C for 120 min.

Experimental design
The effects of the operating conditions on the reaction tests were investigated using a central composite design (CCD) at three levels together with response surface methodology. The CCD consisted of a 2 4 factorial design with 2 center point replicates, resulting in 18 catalytic tests. This method is suitable for fitting a quadratic surface by analyzing the interactions between the parameters with a minimum number of experiments. The central points were used to determine the experimental error and the reproducibility of the results. The independent variables were pH, initial concentration of MB (C MB, 0 ) and hydrogen peroxide ( C H 2 O 2;0 ), and catalyst loading (ω), which were coded to the (− 1,1) interval, where the low and high levels are coded as − 1 and + 1, respectively. The indepeFndent variables with their respective ranges were selected based on the literature as listed in Table 1. The order of the experiments was randomized in order to minimize the effects of the uncontrolled factors. The studied responses were MB degradation, Eq. (1) and COD reduction, Eq. (2). The responses were fitted by multiple regression, and the generated models were used to evaluate the effects of the selected experimental factors according to Eq. (3): where Y is the predicted response, b 0 is the constant coefficient, b i is the linear coefficient, b ij is the interaction coefficient, b ii is the quadratic coefficient, and x i x j are the values of the independent variables. Analysis of variance was used to evaluate the adequacy of the predicted model using Statgraphics Centurion XV.2 software. The quadratic model was adjusted by stepwise selection according to the Fisher-Snedecor test (F-test) in order to remove statistically non-significant variables. The goodness of fit was assessed using the coefficient of determination (R 2 ). Non-significant coefficients were removed from the models (p-value > 0.05).

Results and discussion
Catalyst characterization Figure 1 shows the nitrogen adsorption−desorption isotherm for Nb 2 O 5 /Al 2 O 3 /Al monolith catalyst. The isotherm is of type IV according to the IUPAC classification, indicating the presence of mesopores. It presents a hysteresis loop, which may be attributed to capillary condensation in the pores. The BET surface area of the monolith catalysts was of 43 m 2 g − 1 , with a pore radius of 179 Å and a pore volume of 0.011 cm 3 g − 1 (values referring to the Nb 2 O 5 loading in the monolith). These properties were in good agreement with the results reported by Ohuchi et al. [16] for a Nb 2 O 5 catalyst synthetized through niobic acid calcination. Figure 2 shows the XRD patterns of anodized aluminium monoliths and Nb 2 O 5 /Al 2 O 3 /Al catalyst. The    Fig. 2B). Nb 2 O 5 exhibited different crystalline phases as the temperature increased, showing polyhedral distortion. The TT phase presented a slight distortion of the polyhedra with respect to the hexagonal structure, adopting a pseudo hexagonal structure when the temperature was increased up to 500°C [18]. The Nb 2 O 5 loading of the monolith catalyst was determined by ICP-AES, being of 327 mg Nb 2 O 5 monolith − 1 . The monoliths were subjected to ultrasound testing to determine the adherence the of Nb 2 O 5 layer deposited over anodized aluminum substrate. The coating adherence was found to be excellent (99.8 ± 0.1%). The results are in agreement with those presented by Koyama et al. [19], who deposited a highly stable Nb 2 O 5 layer over anodized aluminum by a sol-gel coating method. Feng et al. [20] also obtained Al 2 O 3 -Nb 2 O 5 composite oxide films with high adherence following a similar preparation procedure to that used in this work. Figure 3 shows SEM micrographs of the surface of Nb 2 O 5 /Al 2 O 3 /Al catalyst and a cross-section of the Nb 2 O 5 deposited layer.
The surface of the Nb 2 O 5 layer was homogeneous, with a thickness of 7 μm, while the thickness of the Al 2 O 3 layer was of 17 μm. The coating exhibited a rough surface, occasionally divided by cracks that went from the surface down to the alumina matrix. This morphology has also been observed by other authors and it may be attributed to shrinkage during the preparation drying stage [21].
The solubility of Nb 2 O 5 in aqueous solutions depends on its crystalline structure, temperature, pH and ionic strength [22]. The Nb 2 O 5 obtained from the alkalization-precipitation of a niobium oxalate solution exhibits an amorphous structure, with a relatively high solubility in water [23]. However, when Nb 2 O 5 is calcined at temperatures above 500°C, a TT-Nb 2 O 5 crystalline phase is generated, decreasing the dissolution rate. Figure 4 shows the dissolution of Nb 2 O 5 after 150 min of reaction at different pH. The samples of the reaction medium were collected and analyzed by ICP-AES to determine the niobium concentration. Figure 4 compares the results for Nb 2 O 5 dissolution (from the Nb 2 O 5 / Al 2 O 3 /Al catalyst) found in this work with those reported by other authors [22,26] as a function of pH (at room temperature). As shown in Fig. 4, the dissolution was significantly lower for TT-Nb 2 O 5 than for the amorphous material. In all cases, an increase in pH yielded an increase in Nb 2 O 5 dissolution for pH 4-12. Based on these results, it was assumed that the reaction only occurred over Nb 2 O 5 surface and the niobia dissolution was negligible. This is in agreement with the results reported by Peiffert et al. for H-Nb 2 O 5 [22].
Timofeev et al. [27] studied the dissolution of Nb 2 O 5 at high temperatures in acid media and reported a solubility of 10 -7.8 M at 150°C and pH 2. Deblonde et al. [26] reported a niobium dissolution of 10 -3.5 M for 56-d aged niobium hydroxide at 25°C and pH 9. Peiffert et al. [22] examined the solubility of anhydrous Nb 2 O 5 at different temperatures and ionic strengths. The highest value reported by the authors for an ionic strength value of 0.1 M (NaClO 4 ) at 25°C was of 6.0 10 − 10 M. Korzhinskaya et al. [23] found a solubility limit value of 8.3 10 − 6 M for Nb 2 O 5 calcined at 550°C, at pH 5 and 0.01 M (Na 2 CO 3 ). It was proposed that niobium oxide dissolution is independent of the ionic strength, between 0.1 and 1 M, and pH 3-8, at room temperature [28].

Mass transport phenomena
The Carberry (Ca) criterion was used to evaluate the external mass transfer limitation in the liquid-solid film according to Eq. (4). The external mass transfer resistance is negligible when Ca < 0.05.
where C i is the concentration of MB or H 2 O 2 (mol m − 3 ) and r obs liq is the observed reaction rate (mol m − 3 s − 1 ). The liquid-solid mass transfer area (a LS ) was calculated from the dimensions of the Al sheets that made up the monoliths as 3834 m 2 m − 3 . The liquid-solid mass transfer coefficient was determined according to the correlation proposed by Butler et al. [29] as k LSMB = 3.85 10 − 5 m s − 1 and k LS H 2 O 2 = 2.34 10 − 4 m s − 1 . The Carberry criterion indicated that the resistance to external mass transfer was negligible for MB and hydrogen peroxide (all the results are shown in Table S4 in Supplementary  Information).
The Weisz-Prater criterion, Eq. (5) was used to estimate the intraparticle diffusion limitation for both methylene blue and hydrogen peroxide.
where D eff is the effective diffusion coefficient (m 2 s − 1 ), L is the thickness of the Nb 2 O 5 layer (m), and ρ c is the apparent density of the catalyst (kg m − 3 ). The effective diffusion coefficient was calculated as follows: The Weisz-Prater criterion indicates that the intraparticle mass transfer limitation for the unknown kinetics is negligible when Φ i < 0.1. The Nb 2 O 5 porosity was ε = 0.22 and the tortuosity was τ = 3 [5]. Intraparticle  Table S4 in the Supplementary Information). This may be attributed to the low effective diffusivity coefficient of methylene blue for the reaction system. Meanwhile, the intraparticle diffusion limitations for H 2 O 2 were negligible.

Effects of the operating conditions on the studied reaction
In this study a central composite design was developed to determine the optimal conditions for MB degradation and COD reduction. The independent variables were: pH, initial concentration of MB (C MB, 0 ) and hydrogen peroxide ( C H 2 O 2;0 ), and catalyst loading (ω). Table 2 shows the complete design matrix together with the values of both responses obtained from the experimental tests.
In order to clarify the contribution of the Fenton-like reaction in MB degradation, the dye was contacted with the H 2 O 2 , without catalyst, and no degradation was observed. When the H 2 O 2 was contacted with the catalyst, no O 2 formation was observed. Both experiments suggest the possible formation of hydroxyl radicals, which would be responsible for the oxidation of MB over Nb 2 O 5 in the presence of H 2 O 2 . The contribution of the Fenton-like reaction was also proposed by Oliveira et al. [13], whose results strongly suggest that the reaction with Nb 2 O 5 catalyst is initialized by the activation of H 2 O 2 to produce •OH radicals.
Under the studied operating conditions and at 150 min of reaction, the minimum MB degradation obtained was 75%, with a COD reduction of 11% (Run 14) or 23% (Run 12). On the other hand, the maximum MB degradation was of 98% with a COD reduction of 79% (Run 1 and 5). Similar results were obtained for both reaction tests at the central point (Run 2 and 11), indicating that the assays were reproducible.
The results of each response were adjusted to a second-order model, using the F-test to remove statistically insignificant variables. Table 3 shows the equations obtained by multiple regression, the percentages of variation of the parameters explained by R 2 , the p-values and the F-values. The coded values for the variables are those presented in Table 1.
The coefficient of determination (R 2 ) obtained for COD reduction and MB degradation was 0.795 and 0.986, respectively. Both second-order models showed p-   values lower than 0.05, ensuring a satisfactory fit of the experimental data with a confidence level greater than 95%. Three-dimensional response surface curves were plotted to understand the interactions between the parameters. Figure 5 shows the response surfaces for MB degradation as a function of two independent variables, and the third and fourth variables were fixed at level "0", as specified in Table 4. Figure 6 shows the response surface for COD reduction. The effects of the operating conditions for MB degradation and COD reduction are discussed below.

a. pH
The influence of pH on MB degradation and COD reduction was studied in the 4-8 pH range. Both responses were favored by lower pH and higher catalyst loadings. As shown in Fig. 5b, an increase in catalyst loading led to a decrease in the pH effect on MB degradation. This behavior has been reported by other authors and it may be attributed to the predominant effect of catalyst loading in the reaction system [10,30].
The pH presents various effects on the reaction system, and it can affect the catalyst surface and/or the reactants, and thus its analysis can sometimes be complex. Changes in pH can alter the adsorption process of the organic molecules on the catalyst surface, which is the first step of the reaction. The isoelectric point of the Nb 2 O 5 coating was pH pzc = 6.8. At pH > pH pzc , the total surface of the Nb 2 O 5 monolith catalyst was negatively charged, whereas at pH < pH pzc the surface had a net positive charge, as shown in the following equations: MB is a cationic dye and its adsorption over the niobia surface by electrostatic interaction is favored at pH > 6.8, as indicated in Fig. 7a. However, there are other possible interactions at low pH between the reactants and the catalyst: a) Lewis acid-base interaction between the axial nitrogen atom of the MB molecule and Nb + 5 sites on the surface (Fig. 7b), b) hydrogen bonding between the nitrogen atom of the central aromatic ring and niobia surface (Fig. 7c), and c) π interaction between the aromatic rings and the niobia surface (Fig. 7d) [25,31].
Oliveira et al. [10,13] proposed that hydrogen peroxide molecules are adsorbed over niobia forming peroxoniobium groups, which are potential oxygen donors. The hydrogen peroxide adsorption on the catalyst surface is also influenced by pH. Wang et al. [32] studied the aerobic oxidation of alcohols by UV radiation using TiO 2 catalyst, observing the formation of peroxo-niobium groups. When a Brønsted acid is present, H + are adsorbed on the catalyst surface, weakening the O-O bond of the peroxo-niobium groups, and thus promoting their heterolytic hydrolysis. In this sense, it can be inferred that under the reaction conditions evaluated in the present work, a decrease in pH led to the regeneration of the active sites on the niobia surface, favoring MB adsorption and its subsequent degradation.
The COD value no depends on the concentration of products and reagents. As shown in Fig. 6, an increase in pH had a negative effect on COD reduction. As stated above, the heterolytic hydrolysis of peroxoniobium groups is promoted at lower pH, favoring H 2 O 2 decomposition, and consequently increasing COD reduction [10].

b. Catalyst concentration
An improvement in MB degradation and COD reduction was observed when the catalyst loading was increased from 2.62 to 7.68 kg Nb 2 O 5 m − 3 . This behavior may be attributed to the increase in total vacant adsorption sites on the Nb 2 O 5 surface, thus favoring the adsorption of MB and H 2 O 2 . It was also observed that an increase in the catalyst loading yielded a greater hydroxyl (•OH) radical generation when hydrogen peroxide was used as oxidant reagent, hence improving dyestuff degradation [32].

c. Initial concentration of reagents
In the present study, the initial dye concentration was between 15 and 45 mg L − 1 . According to the results obtained, MB degradation was no sensitive to initial dye concentration. As shown in Fig. 5A and C, MB degradation was favored by decreasing the initial hydrogen peroxide concentration. This can be explained considering that hydrogen peroxide reagent was in excess in the reaction medium, hence a decrease in H 2 O 2 concentration favored dye adsorption, and subsequently MB degradation.
On the other hand, no statistically significant relationships between COD reduction and initial concentration of the reagents were determined. Optimal hydrogen peroxide concentration for textile wastewater treatment depends on effluent composition and operating conditions [33]. The COD of standard samples was determined to quantify the influence of MB and H 2 O 2 concentrations during COD measurement. The results are presented in Table 4, and they indicate that hydrogen peroxide had a predominant contribution to the COD value. Kang et al. [34] evaluated the interference of hydrogen peroxide on standard COD tests. They proposed the following equation, which can be applied to hydrogen peroxide concentrations in the 0-2000 mg L − 1 range, to correct the effect of the hydrogen peroxide concentration on COD: where C H 2 O 2 is the hydrogen peroxide concentration expressed in mg L − 1 . Figure 8 shows the COD values calculated according to Eq. (9) and those of Table 4, as a function of hydrogen peroxide concentration.
The COD values obtained in the present work for hydrogen peroxide (samples 5 and 6 of Table 4) were in good agreement with those calculated by Eq. (9), indicating that the COD value is mainly determined by H 2 O 2 concentration.
Yahiaoui et al. [35] observed, at a similar MB degradation found in this work (under optimal operating conditions), a substantial increase in the BOD/COD ratio of the effluent after the oxidation process, indicating an increase in its biodegradability. Similar observations were reported by Sapawe et al. [36].
Oliveira et al. [13] proposed a tentative reaction mechanism for the MB degradation over niobia catalyst. The intermediate species were identified by on-line electrospray ionization mass spectrometry, and these results were interpreted with the assistance of the calculation of the Gibbs free energy of the intermediate species. The authors indicated that the reaction is initiated with the activation of H 2 O 2 by niobia via Haber Weiss mechanism, followed by the consecutive hydroxylation of the aromatic ring at position C2, C4 and C5. This is a critical step, as it would simultaneously lead to the formation of hydroquinone or hydroquinone-like intermediates generated by the •OH attack. This would be an unstable key-intermediate that points out the quick and high probability of rupture of both chemical bonds C1-C2 and C5-C6, followed by its oxidation generating a stable species, with m/z = 160.

Kinetic model
Under optimal conditions, the decolorization of MB follows pseudo first-order kinetics according to Eq. (10) [15]: where C MB, 0 is the initial concentration of MB (mg L − 1 ), C MB is the dye concentration at various time intervals, t is the reaction time (min), and k ' is the pseudo first-order rate constant (min − 1 ). By plotting −ln(C MB / C MB, 0 ) vs t, it is possible to calculate the apparent rate constant for MB degradation. Figure 9 presents the experimental and modelled data versus reaction time for MB degradation under optimal conditions. Figure 9 shows that the experimental data were satisfactorily adjusted with Eq. (10) (R 2 = 0.983), demonstrating that the decolorization reaction followed a pseudo first-order kinetics, with an apparent rate constant of0 .03 min − 1 . This value is in good agreement with that reported by Khairnar et al. under similar operating conditions [37].

Catalyst stability test
One of the main advantages of monolith catalysts is the possibility to separate them easily from the reaction medium and use them repeatedly. In this context, the stability of the catalyst activity through consecutive uses is an important factor. To the best of our knowledge, no Multiple consecutive reaction tests under optimal operating conditions with Nb 2 O 5 /Al 2 O 3 /Al catalyst were carried out to analyze the effects of reuse on catalyst activity. The tests were performed using MB solution (in distilled water) and a model textile wastewater, whose composition was: 45 mg L − 1 MB, 5 mg L − 1 NH 4 + , 778 mg L − 1 Na + , 700 mg L − 1 Cl − , 5 mg L − 1 NO 3 − and 150 mg L − 1 SO 4 − 2 [38]. No treatment was performed in the structured catalysts between the tests. Figure 10 shows the dye degradation for ten consecutive tests for MB solution and model textile wastewater treatment. The monolith catalyst reached similar percentages of MB degradation for both cases, and the catalyst activity remained virtually constant after ten consecutive uses. XRD analysis before and after the 10 catalyst stability test were performed in order to explore possible changes in the crystalline structure of the niobia (data shown in Supplementary Information, Fig. S2). The results indicated that the Nb 2 O 5 layer over the anodized aluminum monolith remained unaltered.

Conclusions
The Nb 2 O 5 coating was successfully deposited over an anodized aluminum substrate, showing excellent adherence. The morphology of the surface of the Nb 2 O 5 layer was homogeneous and porous, with a thickness of 7 μm. The dissolution of Nb 2 O 5 in aqueous solution was negligible as a function of pH at room temperature. The