Study of the characteristics and properties of the SiO2/TiO2/Nb2O5 material obtained by the sol–gel process

The SiO2/TiO2/Nb2O5 material was set by the sol–gel method and was characterized by several techniques through thermogravimetric, spectroscopic, and textural analyzes. For the two synthesized materials, the specific surface area was 350.0 and 494.0 m2 g−1 (SiTiNb-A and SiTiNb-B, respectively). An enhance of the crystalline order with the temperature increase of the thermal treatment was observed. Through X-ray Photoelectron Spectroscopy analysis, the binding energy values for the Ti 2p and Nb 3d levels showed the insertion of Ti and Nb atoms in the silica matrix. The Electron Dispersive Spectroscopy analyses also confirmed the high dispersion of the metals presented on the materials surface. The Thermogravimetric Analysis showed weight loss for the of 37.6% (SiTiNb-A) and 29.7% (SiTiNb-B). The presence of the crystalline phases TiO2-anatase and monoclinic-Nb2O5 in the materials was confirmed through the data obtained by association of powder X-ray Diffraction and FT-Raman. Values obtained from optical band-gap aimed the dependence of the oxides concentration and the calcination temperature. Finally, the pyridine adsorption studies have indicated the presence of Lewis and Brønsted acid sites.

Synthesis of the SiO 2 /TiO 2 /Nb 2 O 5 material. The SiO 2 /TiO 2 /Nb 2 O 5 material (designed by SiTiNb) was prepared in two mass proportions, and the samples obtained were SiTiNb-A and SiTiNb-B. The choice theoretical mass percent of TiO 2 and Nb 2 O 5 on the materials was based on other studies of the group, to verify this dependence in the properties, such as acidic properties, specific surface area, optical properties, among others. The synthesis procedure for SiTiNb-A material was: 230.0 mL of a 50% (v/v) ethanol/TEOS solution and 7.0 mL of a 6.0 mol L −1 HCl solution were added to a 500.0 mL reactor. The mixture was stirred for 3 h at 353 K to initiate the pre-hydrolysis of the TEOS. After that, 22.0 mL of titanium (IV) butoxide dissolved in 50.0 mL of ethanol were added drop by drop to the mixture, then more 7.0 mL of a 6.0 mol L −1 HCl solution. The mixture was stirred for 1 h at 353 K. In a third stage, 35.0 g of niobium (V) pentachloride dissolved in 50.0 mL of ethanol were added drop by drop and the mixture was stirred for 30 min at 353 K. After, 15.0 mL of a 6.0 mol L −1 HCl solution were added, and the mixture was stirred at 353 K until the formation of the gel. The resulting mixture was transferred to a beaker glass and heated at 363 K until complete evaporation of the solvent and then heated for 4 h in an oven at 363 K to form the xerogel. The obtained product (xerogel) was precipitated and then dried under vacuum (10 -5 Pa) at 363 K for 6 h. The resulting particles were washed with ethanol in a Soxhlet extractor for 24 h to remove any metallic oxide that was not incorporated into the silica matrix, precursors residues and possible soluble species. Then, SiTiNb-A material was washed with 100.0 mL of a 0.1 mol L −1 HNO 3 solution, followed by some ethanol, ultra-pure water, and ethanol again. Finally, the ternary oxide was dried under vacuum (10 -5 Pa) for 6 h at 363 K and stored.
The same synthesis procedure for SiTiNb-B material was adopted but using the following amounts of reagents: 230.0 mL of a 50% (v/v) ethanol/TEOS solution; 45 Instrumentation and characterization. The amounts of TiO 2 and Nb 2 O 5 in the silica matrix were determined by using Wavelength Dispersive X-ray Fluorescence (WDXRF) on a Bruker AXS equipment (Tokyo, Japan), model S4 Pioneer, operating with Rh tube (E Kα1 = 20.2 keV).
The crystallinity of the thermally treated SiTiNb-A and SiTiNb-B materials was analyzed by powder X-ray Diffraction (XRD). A Bruker AXS D8 Advance instrument (Cu Kα1,α2 radiation, 40.0 kV and 40.0 mA) operating in a Bragg-Brentano θ/θ configuration. The diffraction patterns were collected in a flat geometry with steps of 0.02° and accumulation time of 3 s per step. Finally, the X-ray powder diffraction data were refined following the Rietveld Method with the TOPAS Academic version 5.0 software (Copyright 1992-2012 Alan A. Coelho. Where, for the activation of the program, DLL files are released by Alan).
The analysis of the Specific Surface Area (S BET ) of the SiTiNb-A and SiTiNb-B materials was determined by using the BET (Brauner, Emmett and Teller) multipoint method on a Quantachrome Model Nova 1200e (Boynton Beach, USA) instrument. The samples were previously activated at 393 K in vacuum for 24 h and the samples with granulometry of 150.0 μm ≤ x ≤ 300.0 μm were used. The BJH (Barrett-Joyner-Halenda) method was used to obtain the average pore size and volume.
The Thermogravimetric Analysis was performed on a Shimadzu DTG-60 (Kyoto-Japan). The analyzes, in the range of 293 to 1173 K and the temperature scan rate of 5 K min −1 , were performed with argon flow rate of 50.0 mL min −1 .
Scanning Electron Microscopy (SEM) images were acquired using a JEOL model JSM 6460-LV (Tokyo, Japan) or FEI model Magellan 400 HXR scanning electron microscope at an acceleration voltage of 3.0 kV, 5.0 kV or 10.0 kV according to the analyzed ternary oxide, and 500 × or 1000 × of magnification. For Electron Dispersive Spectroscopy (EDS) analyses, the EDS Noran System Six model 200 (Waltham, USA) was linked to JSM 6460-LV equipment. For the SEM and EDS analyses, the sample was dispersed over double-sided conductive tape on a copper support and coated with gold before the experiment. For this analysis, the materials with a fraction of particle size less than granulometry of 90 μm were used.
The X-ray Photoelectron Spectroscopy (XPS) analysis was performed on the Esca Plus equipment from Omicron Nanotechnology instrument (Germany), in ultra-high vacuum (pressure: 10 -9 mbar), using an Mg X-ray source (Κα = 1253.6 eV). The adjusted emission current was of 16 mA at a voltage of 12.5 kV. Survey spectra were obtained with 50.0 eV analyzer pass energy and 0.5 eV step size. The high-resolution spectra were obtained with 40.0 eV pass energy in the analyzer and 0.08 eV steps. The binding energies were referred to the carbon 1 s of high oriented pyrolytic graphite (HOPG) level, set as 284. 6  www.nature.com/scientificreports/ Solid-state FT-Raman spectra of the thermally treated SiTiNb-A and SiTiNb-B materials were recorded by a Bruker MultiRam spectrometer (Tokyo, Japan) at room temperature with a germanium detector, maintained at liquid nitrogen temperature. For the measurements, 1064 nm Nd-YAG laser line was used with a resolution of 2 cm −1 in the region of 1200-70 cm −1 and 256 scans at a laser power of 100.0 mW. The samples were measured in the hemispheric bore of an aluminum sample holder.
The value determination of optical band-gap of the thermally treated samples was performed through the Kubelka-Munk function (F(R)) of data interpretation from the DRS and using the best function applicable to each curve according to the software. The measurements were performed on a Cary 5000 Varian UV-Vis-NIR Spectrophotometer, with wavelengths between 190 and 950 nm and magnesium oxide as reference.
The acidic properties of the SiTiNb-A and SiTiNb-B materials were verified using pyridine as a probe molecule to detect Lewis and Brønsted acid sites in the samples. Drops of pyridine were added in the samples and they were dried under high vacuum at different temperatures (room temperature, 373 and 423 K). After dried, the samples were analyzed by Infrared Spectroscopy, obtained with pellets containing 12% of the material and 88% of KBr.

Results and discussions
Structural characterization. The XRF analysis was carried to access whether the synthesis procedure was efficient to obtain the oxide mixture 4  Regarding the synthesis, the method showed a lower percentage of titanium oxide and niobium oxide after washing with ethanol in a Soxhlet apparatus with 0.1 mol L −1 HNO 3 solution. These values are lower than expected probably due to the reaction medium pH and other factors, which were not favorable to the complete hydrolysis of titanium (IV) butoxide and niobium (V) pentachloride precursors.
As previously reported by Teixeira et al., it is extremely difficult to obtain SiO 2 /M x O y /N x O y type materials in the desired proportions considering the differences between the hydrolysis kinetics of the precursors in the same reaction media 4 . In addition to that, it can generate soluble species in the washing process, for example. However, the values obtained are satisfactory to employ the SiTiNb-A and SiTiNb-B materials, and the sol-gel process is a method extremely useful and reproductive for the SiO 2 /M x O y /N x O y type materials synthesis 4 .
The Fig. 1 shows the diffractograms obtained after heat treatment at different temperatures for the SiTiNb-A (Fig. 1A) and SiTiNb-B (Fig. 1B) materials. It is observed that both materials, up to a temperature of 873 K do not present enough structural order to be observed peaks in the diffractograms and present low crystallinity, showing a peak for 2θ = 23° (around) due to the amorphous halo typical of glassy silica 30 . However, after 1073 K some peaks begin to appear without having a good definition. At a temperature of 1273 K, can be seen some peaks with greater definition, where the SiTiNb-A sample presents the highest crystallinity at this temperature. Through the values of 2θ of the peaks presented in the diffractograms, when comparing with the ICSD database, it can be identified that beyond the anatase phase of TiO 2 ICSD#92,363 of space group I41/amd Z (141) (a = 3.7710(9) Å ; b = 3.7710(9) Å c = 9.430(2) Å; α = β = γ = 90°) 31 , there is the presence of monoclinic Nb 2 O 5 ICSD#25,750 from space group C 12/m1 (12) (a = 28.51 Å; b = 3.83 Å; c = 17.48 Å; α = 90°; β = 120.8°; γ = 90°) 32 . A halo between 20° and 27° is also observed in 2θ indicating the presence of SiO 2 , as previously mentioned. As it will be seen under by analysis of Electron Dispersive Spectroscopy images (EDS), a high dispersion of metal oxides in the silica matrix is observed and it can be attributed to the strong interactions with siloxane groups forming strong covalent bonds. Also, there is a concentration below 10.0 wt.% of TiO 2 in both materials. This high dispersion and low concentration of TiO 2 in the silica matrix decreases the mobility TiO 2 in the silica matrix, making it difficult in the formation of Anatase phase or Rutile at lower temperatures.
The Tables 1 and 2 summarize the crystallographic data obtained using the cell parameter values, crystallite size referring to TiO 2 in the anatase phase and monoclinic-Nb 2 O 5 phase obtained with the refinement by the Rietveld Method. In the case of SiTiNb samples, by virtue of their low crystallinity, it was only possible to perform the refinement for the diffractograms of the samples calcined from 1073 K. The Gof (Good of fitness) and Rwp (weighted profile R-factor) values were lower and better than those obtained in the refinement of samples containing niobium, showing a good approximation of the theoretical and experimental diffractograms.
It can be observed that, with the increase in the calcination temperature, there is an increment in the size of the crystallite, while the volume of the unit cell undergoes a small reduction, which is characteristic of the grow in the structural ordering of the sample. However, in the SiTiNb-B sample, there is a slight gain in the volume of the unit cell of the sample calcined at 1073 K in relation to that calcined at 1273 K, indicating a higher amount of anatase sites in relation to monoclinic niobium oxide may be the cause of this reduction (Tables 1, 2, 3).
According to refinement, the relative percentage between the anatase and monoclinic phases varies as specified by Table 3. It is possible to observe that the calcination temperature increase favors the anatase-TiO 2 phase and decreases the percentage of the monoclinic-Nb 2 O 5 phase. This behavior occurs in the same proportions in the two samples, demonstrating that it is only an effect of temperature and it is not related to the amount of metal oxide in the matrix of each material.
The BET analyses showed that the specific surface area was 494.0 m 2 g −1 for SiTiNb-A and 350.0 m 2 g −1 for SiTiNb-B. The BJH method revealed an average pore volume of 0.272 cm −3 g −1 for SiTiNb-A and 0.194 cm −3 g −1 for SiTiNb-B; and the average pore size of 11.0 Å for SiTiNb-A and 11.1 Å for SiTiNb-B, which indicates that these ternary oxides are microporous. These values are important, as it shows that the active sites are accessible in the SiTiNb-A and SiTiNb-B materials. www.nature.com/scientificreports/ It can be seen by the adsorption and desorption curves (Fig. 2) for both materials, that the isotherms are  www.nature.com/scientificreports/ type I(b) 33 , presenting a characteristic curve of microporous materials with a small external surface area 34,35 . In such cases, it is common for adsorption to occur on the accessible pore volume and not on the internal surface area. There is a small separation between the adsorption and desorption curves, pointing out a possible H4-type hysteresis 33 , which is normally associated with adsorption in micropores and characteristic of non-linear slits 36 . However, the BET analyses demonstrate a lower surface area and pore volume for SiTiNb-B, showing that the concentration of TiO 2 and Nb 2 O 5 oxides in the silica matrix had a greater influence on the parameters (surface area and pore volume) and probably hardens the structure due to the formation of a dense layer and/or the coalescence of pores 7,8 .
The thermal stability of SiTiNb-A and SiTiNb-B materials is evaluated by Thermogravimetric Analysis (TGA-DTA) (Fig. 3). For the SiTiNb-A material (Fig. 3A), in the temperature variation to approximately 383 K, an endothermic event (exo down) is observed with a mass loss of 21%, related to water desorption. From 383 to 528 K (exothermic event, exo up), it was found a mass loss of approximately 8%, due to the dehydroxylation of the groups -SiOH, -TiOH or -NbOH in virtue of structural water loss 37,38 . Above the temperature of 528 K, there was a change in the line referring to DTA, which may indicate an event where there is a change in energy generating a mass loss of approximately 4% (endothermic event, exo down). This event can also be seen in Raman's analysis which will be discussed later. In the range between 673 and 1093 K there is a mass loss also of  Finally, there is a small loss (0.6%) ascribed, possibly, to organic matter aggregated or encapsulated in the pores of the sample 40 , totaling a mass loss of 37.6%. For SiTiNb-B material (Fig. 3B), the mass losses are similar, totaling 29.7%. Up to 408 K a mass loss of 15.4% is observed (endothermic event, exo down), related to water desorption 37,38 . Between 408 and 523 K the mass loss is approximately 5% (exothermic event, exo up) and in the range of 523 K to 698 K the mass loss also corresponds to 5% (endothermic event, exo down) 38,39 . At 698 K at the end of the process, the mass loss is 4% [38][39][40] .
As seen in Fig. 4, the Scanning Electron Microscopy (SEM) for SiTiNb-A (Fig. 4A) and SiTiNb-B (Fig. 4B) materials shows the morphology of the particles does not present a uniform size and it presents irregular shape and no-spherical particles. They have been characterized by a flat surface with a rough structure. This characteristic is quite usual for mixed oxides obtained by the sol-gel process 3,4 . Note that the particle size distribution is quite heterogeneous, which is characteristic for mixed oxides obtained by the sol-gel process; and the distribution cannot be estimated by SEM analysis. But there are particles in the vast majority between 10 and 90 µm. Some particles are even larger than 100 µm, but as rods form, they pass through the mesh of the sieve that was used for particles smaller than 90 µm.
The Fig. 5 shows the scanning by EDS analysis of Si (Fig. 5B), Ti (Fig. 5C) and Nb (Fig. 5D) for SiTiNb-A material. The EDS shows the TiO 2 or Nb 2 O 5 particles are dispersed without phase segregation or formation of islands at the silica matrix, which was a desired feature. In addition, at the magnification level used (1000x), the SiTiNb-A material seems to be homogeneous with high uniformity on its surface. This dispersion is very important because increases the number of acidic sites on the surface or pores of SiTiNb-A material, and it can be attributed to the strong interactions with siloxane groups of the silica matrix forming strong covalent bonds as the -Si-O-Ti-OH, -Si-O-Nb-OH types, as observed by XRD analyzes.
The percentage values on the surface of the SiTiNb-A material for the elements Si, Ti and Nb were determined from the 3-point analysis by EDS, which were homogeneous, corroborating with the XRF analysis. It was observed through the SEM and EDS analyzes that the SiTiNb-B material presented the same behavior, so the results were not presented.
The XPS analyses were performed to investigate the chemical states of the elements in the SiTiNb-A and SiTiNb-B materials. The survey spectra (Figs. 6A, 7A) show the presence of O, Ti, Nb and Si elements, which confirm the mixture of SiO 2 , TiO 2 and Nb 2 O 5 oxides on the samples surface, also observed in EDS analyses. It was also observed in spectra the presence of residual carbon, accordingly to Barr and Seal 41 is common on XPS      8A) and SiTiNb-B (Fig. 8B) materials, there was an appearance of crystalline phase at 1073 K, these bands being present in the sample without calcining and in those calcined at temperatures below 1073 K. With a higher temperature of calcination, this ordering increases and it can also be noticed in the medium and long distance as data observed in the XRD analyzes. According to the Raman spectra of thermally treated samples and the spectroscopic factor group analysis, the anatase phase presented six vibrational modes active in Raman (D 4h   19 ).
(A)    www.nature.com/scientificreports/ was less definition of the bands in relation to sample A, however, the well-defined peak, at 144 cm −1 , referring to the anatase, appears at 873 K. Note that in the 150-100 cm −1 region there was no overlapping of the bands referring to anatase and niobia. The optical band-gap values were obtained using the Kubelka-Munk (F (R)) function in the Diffuse Reflectance data of the heat-treated samples (Fig. 9) and are shown in Table 6. Nb 2 O 5 is n-type semiconductor with a band-gap of about 3.4 eV 51 , and that was dependent on oxygen stoichiometry in its structure, varying its bandgap from 3.2 to 4.0 eV. In general, Nb 2 O 5 had a higher conduction band than TiO 2 being indicated to obtain open-circuit voltage and efficiency in the conversion of photons. The most thermodynamically stable form of Nb 2 O 5 is the monoclinic arrangement presented in the synthesized samples 52 . For the TiO 2 anatase band-gap is observed in the 3.23 eV region 53 ; and this transition occurring in anatase can be described as an absorption of the valence band (essentially 2p filled O 2-) to the conduction band (essentially 3d voids of the Ti).
The samples which were not calcinated had a band-gap around 3.3 eV and when the appearance of the TiO 2 anatase and Nb 2 O 5 monoclinic phases begun, there was a slight increase in the band-gap followed by its reduction after temperatures of 1073 K. Whereas the monoclinic phase stabilizes, it is obtained a band-gap value of 3.1 eV in 1273 K. Thus, the transition from band 2p (full) of oxygen (O 2-) to band 4d (empty) of Nb 5+ should occur. It  www.nature.com/scientificreports/ was observed in the spectrum of SiTiNb-A phase a tooth formation (electronic artifact) that masks the proper measurements, requiring deeper analysis to determine minor variations in the band-gap values. Table 6 shows that the band-gap values did not vary significantly with the different percentages of the crystalline phases. However, in Tables 1 and 2 were shown a variation in the crystallites size of the crystalline phases and a smooth "red-shift" in the energy of band-gap, both in the heat-treated samples.
The acid properties of the SiTiNb-A (Fig. 10A) and SiTiNb-B (Fig. 10B) materials were studied using pyridine adsorbed as probe molecule, which is the most used in the literature 54 . The characteristic bands for Lewis and Brønsted acid sites can be observed in both materials: The Lewis acid sites are due to the titanium or niobium ions coordinatively unsaturated [55][56][57][58] , and the Brønsted acid sites are a contribution of Ti-OH or Nb-OH groups, for example 11,59-62 . Table 6. Optical band-gap of SiTiNb-A and SiTiNb-B from the function of Kubelka-Munk.

Sample
Optical band-gap (eV) Sample Optical band-gap (eV) A  3,3  SiTiNb-B  3,3   SiTiNb-A-473  3,3  SiTiNb-B-473  3,3   SiTiNb-A-673  3,3  SiTiNb-B-673  3,4   SiTiNb-A-873  3,5  SiTiNb-B-873  3,4   SiTiNb-A-1073  3,1  SiTiNb-B-1073 3,2   SiTiNb-A-1273  3,1  SiTiNb-B- www.nature.com/scientificreports/ The weak band at 1575 cm −1 is due to the vibrational mode 8a of the pyridine physically adsorbed by weak forces such as van der Waals. Thus, the pyridine molecule exhibits low interaction with the material, and this band disappears whereas the temperature increases 63,64 . The bands at 1598 and 1446 or 1447 cm −1 , certainly, are due to the 8a and 19b modes of pyridine bonded to the free silanol groups present at the surface of the materials. The -SiOH group of silica matrix by hydrogen bonds seen in the pure SiTiNb ternary oxide treated at room temperature disappears whereas the temperature increases 65 . The band at 1490 cm −1 possibly corresponds to the vibrational mode 19a. A band at 1544 cm −1 is attributed to the vibrational mode 19b of the pyridine molecule bound to the Brønsted acid sites (-TiOH or -NbOH groups), and this band is not observed for the pure SiO 2 66 . The 19b mode of pyridine bound to Lewis acid sites is observed as a small shoulder at approximately 1458 cm −1 that disappears after heat treatment at 423 K. Finally, the weak band detected at 1641 or 1639 cm −1 characterizes the presence of Brønsted acid sites, that decreases while increasing temperature 67 .

SiTiNb-
In comparison to the SiO 2 /TiO 2 /Sb 2 O 5 (SiTiSb) material, already described by LaDANM Group 4 , the behavior is different in relation to the SiTiNb material, in this study. It can be reported that in the SiTiSb-A material the presence of the anatase phase is noticed in the sample treated at 473 K. However, in the system with niobium, SiTiNb material, it was only observed the presence of the anatase phase with the appearance of the crystalline oxide phase of monoclinic niobium at 1273 K. This indicates that the SiO 2 /Nb 2 O 5 network binds more strongly to Ti than the SiO 2 /Sb 2 O 5 network.
The different structural characteristics and properties showed for the SiTiNb materials with the presence of the Lewis and Brønsted acid sites on its surface made them interesting for use involving application studies, such as adsorbent for metal ions, in photocatalysis, etc. In addition, the band-gap values obtained for both SiTiNb-A and SiTiNb-B show they can be probably used as n-type semiconductors.

Conclusions
The SiTiNb materials obtained by the sol-gel method, showed a high dispersion of metal oxides at the silica matrix, showed no phase segregation, and presented significant Specific Surface Areas (S BET ). The XPS analysis indicated the insertion of Ti and Nb atoms in the silica matrix and that antimony is in its higher oxidation state. These materials presented good thermal stability as observed by DRX and TGA-DTA data. Crystalline phases in the ternary oxides were found only after heating at 1073 K and confirmed by the optical band-gap depend on the oxides concentration and the calcination temperature. The presence of the Lewis and Brønsted acid sites on its surface was confirmed by pyridine test. Due to the characteristics presented of these materials, they are (the most) promising for testing as electrochemical sensors, adsorbent for metal ions and other species in effluents, heterogeneous catalysts and photocatalysis, for instance.