Sodium Tungsten Oxide Bronze Nanowires Bundles in Adsorption of Methylene Blue Dye under UV and Visible Light Exposure

This paper describes the analysis and characterization of NayWOx bronze nanowires bundles and evaluation of their effective adsorption of methylene blue dye (MB). The Na-doped WOx bronze nanowires bundles were first synthesized via a simple solvothermal method, which were then fully characterized by using different techniques including TEM, XRD, XPS and UV-Vis, to validate the successful Na+ insertion into the WOx framework. The adsorption activities of the resulting NayWOx bronze nanowires bundles, compared with the undoped WOx form, were investigated by evaluating the adsorption effect on methylene blue under both UV and visible light irradiations. An enhanced adsorption performance of the Na-doped WOx bronze samples was recorded, which demonstrated a 90% of removal efficiency of the MB under different conditions (dark, visible and UV light). Moreover, the NayWOx bronze samples also offered a 4 times better kinetic rate of MB removal than the plain WOx nanowires.


Introduction
Water contamination is one of the key environment issues facing our globe, due to the rapid developments of industrialization and growth in population. While we are struggling to cope with the high demand for water resources, an increasing amount of wastewater is generated, which undermines the quality of precious water resources [1]. According to a report by the World Health Organisation (WHO), huge areas of the world are suffering from poor water quality, involving about 844 million people [2]. Apart from discovering new clean water resources, wastewater has been considered as one of largest possible alterative resources that can be treated and reused, to offset the high demands for existing water resources [3]. In general, wastewater is classified according to the source of contaminations containing different types of pollutants, such as aliphatic hydrocarbons, grease, heavy metals, organic compounds, etc. [2]. Among these pollutants, dyes are one of the most common organic compounds found in wastewater, which are widely used in numerous industry sectors, such as paper, leather, plastics, pharmaceutical, cosmetics, textiles, dyestuffs, etc. [4]. Dyes are generally not biodegradable and can persist in the environment for a long time, hence generating direct impacts on human health and the environment [5]. 2 of 16 Amongst many wastewater treatment technologies developed to date, semiconductorbased adsorption and photocatalysis are the most promising techniques, due to their high efficiency, low-cost, environmentally friendly nature, and sustainability. These techniques have the potential for the removal of both harmful bacteria and organic pollutants by converting the pollutants into comparatively benign water and carbon dioxide [6,7]. In detail, the light absorption generates charge separation by photo-exciting electrons from the valence band to the conduction band, forming an electron/hole pair in the photocatalyst materials. The positive holes produce OH radicals by oxidizing OH − anions, which could react with pollutants and convert them into a benign substance. Semiconductor-based adsorbent/photocatalysts are water insoluble, thus they can be either grown on substrate such as stainless steel wire [8] and polymeric fiber [9], or coated as a smart material onto supporting substrates [10], to minimize the loss during wastewater treatment. By adopting these approaches, the semiconductor-based adsorbent/photocatalysts could also easily removed from the reaction medium after adsorption/photocatalytic processes. However, current adsorbent/photocatalysts still cannot meet all the desired requirements, including high removal efficiency, full spectrum utilization of sunlight, large specific surface and good recyclability [7]. For several decades, the search and development of cost-effective and highly removal efficient for environment remediation has remained a huge challenge.
To improve the adsorption activity of WO 3 , two techniques can be applied [27][28][29][30]: (1) the synthesis of nanostructured WO 3 with tailored morphology; and (2) the modification of WO 3 by metal, non-metal and organic materials, as these could tune the band structures of the materials. An open channel cavity structure in the blue tungsten oxide (sub-stoichiometric structures, x < 3) or in the tungsten oxide bronze-type compound with a general formula of M y WO 3 (where M = Rb + , H + , Na + , NH 4 + , etc.) has the potential to drive efficient adsorption and photocatalytic reactions [31]. Furthermore, these blue oxides can be activated by both visible and UV light, which is a highly desired feature. So far, adsorption performance of tungsten oxide bronze still lacks detailed studies, with very few reports involving Na y WO 3 [26], Cs y WO 3 [26], etc. being published. Herein, we first report the synthesis of both novel blue sub-stoichiometric tungsten oxide (WO x , x < 3) and Na-doped tungsten oxide bronze (Na y WO x ) structures, and then investigate their adsorption performance towards the removal of methylene blue (MB) under both UV and visible light irradiations. This study allows us to gain a deep insight into tungsten oxide-based adsorbent, offering guidance for the development of new adsorption technology.

Materials
All raw materials and solvent were purchased from Sigma and were used without further purification. Tungsten hexachloride (WCl 6 , ≥99.9% trace metal basis) and sodium chloride (NaCl, 99%, AR grade) were used as the precursors for WO x . Cyclohexanol (C 6 H 11 OH, ReagentPlus 99%) was used as the solvent. Methylene blue (C 16 H 18 CIN 3 S·xH 2 O, ≥82%) dye was used as the model pollutants for the adsorption study because their concentration can be easily monitored using a spectrometer.

Sample Preparation
The as-prepared Na-doped tungsten oxide (Na y WO x ) nanowires bundles were synthesized by using a solvothermal method. In detail, a certain amount of WCl 6 was dissolved in cyclohexanol to form a 4 mM solution. A total of 50 mL of the well-dissolved solution was poured into a 120 mL Teflon-lined stainless steel autoclave. Solvothermal synthesis was conducted at 200 • C for 6 h in an electric oven. For the Na-doped WO x samples, 5 different concentrations of 10 mL NaCl pre-solution were prepared using distilled water as the solvent. The concentration of NaCl pre-solution was prepared based on molar NaCl and WCl 6 ratios of 1:16, 1:12, 1:8 and 1:4, respectively. Then, 1 mL of NaCl pre-solution was added into the well-dissolved 49 mL WCl 6 /cyclohexanol solution, which was gently stirred and then transferred into the autoclave. The final concentration of the WCl 6 was still 4 mM. The autoclave was sealed and heated at 200 • C for 6 h. After the reaction, all as-synthesized samples were washed several times using distilled water, ethanol and acetone, respectively. The particles were obtained by centrifugation and dried overnight in an oven at 80 • C.

Characterization
Structures and morphologies of the as-prepared samples were characterized by X-ray diffraction (XRD, Brunker D8 Advance diffractometer with a Cu Kα radiation, operated at 40 kV−40 mA), Scanning Electron Microscopy (SEM, Philips XL-30, operated at 20 kV), Transmission Electron Microscopy (TEM, JEOL-2100, operated at 200 kV) and X-ray Photoelectron spectroscopy (XPS, VG ESCALab Mark II spectrometer with a non-mono-chromatic Al-anode X-ray source (1486.6 eV)) operated at a 12 kV anode potential and a 20 mA filament emission current. Fourier-Transform Infrared spectroscopy (FTIR, Perkin Elmer Spectrum One) was used to scan from 4000 cm −1 to 400 cm −1 and the number of scans was 16. The resolution of the FTIR measurement was 1 cm −1 . Sputter coating (gold, 5 nm) was used to prepare the SEM samples, to avoid charging. The BET surface areas of the samples were measured by N 2 sorption at 77 K using a Quantachrome autosorb iQ2 ASiQwin apparatus via the conventional volumetric technique. Before the analysis, all samples were degassed at 200 • C for 3 h under a vacuum. The pore size distribution (PSD) was determined using the non-local density functional theory (NLDFT) method.

Adsorption Performance
The removal of MB by Na y WO x samples was evaluated under the condition of dark, visible and UV light irradiation. In a typical experiment, 0.01 mg of WO x based samples was dispersed in 100 mL of MB solution with a concentration of 3.5 mg/L. The removal of MB by these samples via physical adsorption was performed under dark conditions. The solution was sampled at 15 min time intervals for a total of 120 min. The adsorption performance of these particles was assessed under irradiation of visible light (λ = 555 nm) and UV light (λ= 254 nm). The MB removal efficiency of these particles was assessed under different optical excitation sources. Three pieces of fluorescent lamps (Phillips TL-D 18W/54-765) were wrapped with UV filters to provide the visible light (λ = 555 ± 1 nm), whereas a UV mercury lamp (Phillips TUV T8 F17) was used to provide the UV light (λ = 254 ± 1 nm) in this study. The distance between the light source and solution was kept at 10 cm.
The intrinsic absorbance of MB solution at 663.9 nm was used to calculate the concentration of MB dye. The absorbance of MB solution was recorded using a Varian Carry 50 UV-Visible spectrophotometer. The absorbance of solution was measured from 400 nm to 800 nm, at a scanning step of 1 nm with a scan speed of 10 nm s −1 . The removal efficiency (RE) of MB by the particles was calculated according to Equation (1) [32]. where C 0 and C t are the absorbance intensity of the MB dye solution at the initial and a given time interval, respectively. The reusability evaluation of Na y WO x bronze samples on the removal of MB dye was carried out under UV light irradiation and under dark conditions for 4 cycles. The irradiation time for each cycle was 120 min. The particles were collected by filtration (filter papers, Sigma Aldrich, 1441-090) and rinsed with deionized water to remove the MB residual, prior to the subsequent cycles of the MB removal test.  Figure 1b). The block bundles of Na y WO x bronze sample are shorter in length and wider in diameter than those of the pure WO x . By increasing the Na-dopant amounts to 1:12, 1:8 and 1:4, mixtures of nanorods, nanoplates and agglomerated bundled structures were observed, as shown in Figure 1c-e, respectively. The size of nanoplates is larger with increased Na-contents inside the framework. Further morphological investigations from TEM show that W 18 O 49 nanowires consisted of ultrathin nanowires of only ca. 2-5 nm in diameter and up to 2 µm in length, which are self-assembled into larger bundles, as seen as individual fibers under the SEM, due to the low resolution limit (Figure 1f).

Results and Discussion
The XRD pattern of as-prepared WO x ( Figure 2) matches with the spectral pattern of monoclinic W 18 O 49 (PDF 01-084-1516), with lattice constants of a = 18.32 Å, b = 3.78 Å, c = 14.03 Å and β = 115.21 • . Two main diffraction peaks at 23.2 • and 47.5 • can be assigned to the (010) and (020) planes, respectively. The strongest intensity of the (010) plane suggests that the nanowires are preferably grown along the <010> direction, which also agrees with the HR-TEM images and SAED patterns (Figure 1f,k). Higher Na-doped WO x samples (1:12, 1:8 and 1:4) not only caused changes in the morphology, but also slightly modified their crystalline structure. The XRD patterns of 1:12, 1:8 and 1:4 Na-doped WO x samples match with standard patterns of hexagonal Na y WO 3 (PDF 00-046-0174). The lattice constants of the hexagonal phase Na y WO 3 (222) were also visible. It is possible that a small amount of nanoplates and larger nanorods could co-exist with the nanowires, which exhibited as different phases in the 1:16 sample.
An enlarged view of the XRD pattern of the bronze samples suggested that the peak position of the (002) plane (located range from 23.08 • to 23.32 • ) was shifted to a lower diffraction angle for higher Na amounts. Based on Bragg's law, the observed peak shift can be related to the increase in lattice parameters, which confirms the intercalation of different Na contents into the WO 3 layers and formed the bronze-type structure. A summary of the intensity ratios of each sample is presented in Table 1. pers, Sigma Aldrich, 1441-090) and rinsed with deionized water to remove the MB res ual, prior to the subsequent cycles of the MB removal test.  ated bundled structures were observed, as shown in Figure 1c-e, respectively. The si nanoplates is larger with increased Na-contents inside the framework. Further mor logical investigations from TEM show that W18O49 nanowires consisted of ultrathin owires of only ca. 2-5 nm in diameter and up to 2 µm in length, which are self-assem into larger bundles, as seen as individual fibers under the SEM, due to the low resolu limit (Figure 1f). The XRD pattern of as-prepared WOx ( Figure 2) matches with the spectral patte monoclinic W18O49 (PDF 01-084-1516), with lattice constants of a = 18.32 Å, b = 3.78 Å 14.03 Å and β = 115.21°. Two main diffraction peaks at 23.2° and 47.5° can be assigne the (010) and (020) planes, respectively. The strongest intensity of the (010) plane sug that the nanowires are preferably grown along the <010> direction, which also agrees the HR-TEM images and SAED patterns (Figure 1f and 1k). Higher Na-doped WOx ples (1:12, 1:8 and 1:4) not only caused changes in the morphology, but also slightly m ified their crystalline structure. The XRD patterns of 1:12, 1:8 and 1:4 Na-doped WOx ples match with standard patterns of hexagonal NayWO3 (PDF 00-046-0174). The la constants of the hexagonal phase NayWO3 include a = 7.  The intensity ratios of (002)/(200) decreased upon increasing the Na + concentrations. This correlation suggests that the incorporation of Na ions into the hexagonal tunnel of WO 3 has changed the original plane stacking characteristics; it either reduced the number of (002) planes, i.e., larger nanorods and nanoplates, or enhancing the growth of other planes. Meanwhile, the Full-width at half-maximum (FWHM) of the (002) plane also showed greater values, whilst the (200) plane presented smaller values with increased Na:W molar ratios, i.e., corresponding to smaller size on the (002) growth direction and larger size on (200) growth direction. These data match very well with the SEM evidence that nanorods and nanoplates became bigger at higher Na + concentrations. In addition, HR-TEM lattice and SAED pattern images of pure W 18 O 49 and the 1:16 doped samples (Figure 1k,l) confirmed that the lattice spacing was 0.379 nm, which matched well with the d value of the (010) plane, as confirmed by the XRD results. However, the d values and SAED patterns of higher Na-concentration samples were different, as presented in  XPS was used to analyze the surface chemistry of the pure W 18 O 49 and Na y WO 3 bronze samples. The results of survey spectra of Na 1s, W 4f and O 1s were presented in Figure 3. The spectra clearly showed that elements of W, O and Na existed in all samples, and the atomic concentration values of each element are summarized in Table 2. eV (peak 1) and 532.4 eV (peak 2) were observed [34]. The peak at the Na 1s region was observed at about 1072 eV [35].  [37] inside the lattice of WO3. We suggested that the Na + was inserted into the hexagonal 248 tunnels of the WO3 framework, as confirmed from the XRD pattern. As a result, the oc-  For pure W 18 O 49 , the two spin-orbit doublets that were attributed to the two different oxidation states of W atoms W 6+ and W 5+ were observed [33]. The main doubles for W 4f 5/2 and W 4f 7/2 with binding energy at 38.56 and 36.43 eV, respectively, were assigned to the W 6+ oxidation state. The binging energies of W 4f 7/2 at 37.04 eV and W 4f 7/2 at 34.91 eV could be allocated as the W 5+ oxidation state. For the O 1s spectra, two main peaks at 530.8 eV (peak 1) and 532.4 eV (peak 2) were observed [34]. The peak at lower binding The peak at the Na 1s region was observed at about 1072 eV [35]. By increasing the amounts of Na dopant, W 4f and O 1s peak positions of the bronze samples shifted toward the lower binding energy, compared with the pristine W 18 O 49 structure. The ratios of W 5+ /W 6+ and O 1s peak 1/peak 2, which corresponded to the ratios of the fitted peak areas, are also shown in Table 2. Na atomic contents in the bronze samples of 1:4, 1:8, 1:12 and 1:16 were 6.8%, 6.1%, 4% and 3.2%, respectively. The atomic contents of W and O were increased first and then decreased, while increasing the concentration of Na + inside the structures. The increase of W 5+ /W 6+ and the O 1s peak 1/peak 2 atomic ratios were indirectly related to the increased oxygen vacancies (V o ) inside the framework structures. The results suggested that the V o could be increased with the increase of Na dopant amounts.

Structure and Morphology
Due to the large ionic radius of Na + (0.98 Å) [36], it could be difficult for Na + to replace either W 5+ or W 6+ (which have smaller atomic radii of about 0.65 and 0.68 Å, respectively) [37] inside the lattice of WO 3 . We suggested that the Na + was inserted into the hexagonal tunnels of the WO 3 framework, as confirmed from the XRD pattern. As a result, the occurrence of Na + in the bronzes could enhance the electron transfer and oxygen mobility of the samples, as confirmed by the altered oxidation states obtained from the XPS results. The peak shifts of the W 4f and O 1s in the bronze samples offered additional evidence of increased defects and weaker W-O bonding inside the structure, resulting in lattice relaxation and structure modification in bronze samples. The energy of optical band gap of all samples was calculated using the Tauc-equation (Equation (2)) [38] (αhυ) 1/m = k (hυ − E g ) (2) where (E g ) is the energy of optical band gap, k is a constant and m = 2 in the case of an indirect energy gap. (αhυ) 0.5 was plotted versus hυ and the linear portion of the plot was extrapolated to the ordinate. As shown in Table 3 Figure S1, all samples exhibited type IV N 2 adsorption/desorption isotherms. The hysteresis loop was observed at relative pressure of above 0.4 p/p o due to the capillary condensation taking place in mesopores [39]. NLDFT Pore size distributions (shown in Figure S2), which also confirmed that all samples were predominantly mesoporous with an average pore sizes of 2.8 nm. However, larger pores also formed at 5-15 nm. The proportion of mesopores within the 5-15 nm range increased in the Na y WO 3 samples. BET surface areas and pore volumes of the measured samples are also listed in

Adsorption Activity
The absorption spectra of MB dye (3.5 mg/L) for all samples that were tested under the conditions of dark, visible and UV light are presented in Figures 4-6, respectively. The characteristic absorbance peak of the MB solution at 663.9 nm was clearly observed in all samples at t = 0 min. The peaks decreased dramatically within the first 30 min for all samples under all testing conditions. As no optical excitation was provided, the decay of 663.9 nm peak indicated that the MB removal at this stage could be assigned to adsorption rather than to photo-degradation. After 30 min, there was an obvious blue shift of the maximum absorbance peak from 663.9 nm to 576.9 nm. This blue shift was attributed to the formation of trimer (MB + ) 3 on the surface of W 18 O 49 or Na y WO 3 bundles, due to the electrostatic attraction between the oxide samples and MB molecules [32,40].  It is also noted that the removal efficiency after 120 min without using any samples was also recorded as 0.19% under dark conditions, 10.32% under visible conditions and 28.91% for UV conditions (Figure 7d), for comparison. The result suggested that some of the MB dye could be adsorbed on the internal wall of the glass vial.
In order to evaluate the effect of light sources on the removal of MB dye, pair-t tests (α = 0.05) were performed by comparing the removal efficiencies under various conditions: dark vs. UV light, dark vs. visible light and UV light vs. visible light. Table 4 shows that there was no difference for the removal efficiency either between UV and visible light irradiation or between dark conditions and visible light irradiation. However, there was a difference on the removal efficiency between UV light irradiation and dark conditions. The higher removal efficiency by the samples under UV light was due to the synergy effect of adsorption (main removal mechanism) and degradation of MB molecules caused by the UV light. Table 4. Pair-t test for UV light vs. visible light, visible light vs. dark conditions, and UV light vs. dark conditions. The null hypothesis (Ho) and alternative hypothesis (Ha) were stated as follow: Ho: µ1 − µ2 = 0; Ha: µ1 − µ2 = 0, where µ is removal efficiency. The confidence interval was at 95% (α = 0.05). The adsorption of MB on W 18 O 49 and Na y WO 3 samples was further confirmed by FTIR analysis, as depicted in Figure S3. According to the FTIR spectrum of MB, the peaks at 3420 cm −1 and 2928 cm −1 were attributed to the NH/−OH overlapped stretching vibration, and symmetrical stretching of C-H, respectively. The strong peak at 1599 cm −1 belonged to the stretching band of C=O and C-N from amine groups, while the band at 1489 cm −1 came from a symmetrical stretching band of -COOH. The peak at 1394 cm −1 and 882 cm −1 indicated the bending band of N-H and C-N from the amine group, respectively [41]. In the FTIR spectra of W 18 O 49 and Na y WO 3 (1:8) samples, the wide band at 3436 cm −1 was assigned to the stretching vibration of O-H groups, while the peak at 1619 cm -1 came from the W-O-H vibration. The wide band ranging from 500 cm −1 to 1000 cm −1 was related to the W-O vibration mode. For instance, the W-O stretching mode resulted in the spectral region of 600-659 cm −1 , and the strong peak at 815 cm −1 could be attributed to the W-O-W stretching mode [42]. When the Na-dopant was introduced into the WO 3 lattice, the additional peak at 1464 cm −1 was due to the vibration of Na-W-O bonding. After the adsorption of MB onto the surface of W 18 O 49 or Na y WO 3 samples, the amine functional group (R-NH 3 ) was clearly observed as peaks at 1394 cm −1 and 1349 cm −1 [43]. As there are no significant shifts of the R-NH 3 peak for W 18 O 49 -MB and Na y WO 3 -MB, we suggested that the binding of MB on W 18 O 49 and Na y WO 3 may not have arisen from the formation of the hydrogen bond.

N
The surface charges of the samples were measured using zeta potential. As shown in Figure 8, the W 18 O 49 and Na y WO 3 particles were negatively charged in distilled water. The zeta potential of all samples varied from −43.2 mV to −24.6 mV, indicating that W 18 Figure 9 showed the reusability of Na y WO 3 (1:8) in the removal of MB dye under dark conditions and UV light irradiation for 4 consecutive cycles. Under dark conditions, the removal percentage slightly dropped from 93.9% to 77.9% in the second cycle, and dramatically decreased to 27.5% and 9.7% at the third and the fourth cycle, respectively. A similar trend of removal efficiency was found in the UV light studies; however, it was higher than those obtained in the dark. For instance, the removal efficiency dropped from 94.7% to 92.5%, 40.9% and 12.9% at the second, third and fourth cycle, respectively. The improved efficiency by 0.8%, 14.6%, 13.4% and 3.2% for first, second, third and fourth cycle, respectively obtained after the introduction of UV irradiation could be attributed to the adsorption process initiated by the UV light.
of adsorption (main removal mechanism) and degradation of MB molecules caused by 328 the UV light. 329 The adsorption of MB on W18O49 and NayWO3 samples was further confirmed by 330 FTIR analysis, as depicted in Figure S3. According to the FTIR spectrum of MB, the peaks 331 at 3420 cm −1 and 2928 cm −1 were attributed to the NH/−OH overlapped stretching vibra-332 tion, and symmetrical stretching of C-H, respectively. The strong peak at 1599 cm −1 be-333 longed to the stretching band of C=O and C-N from amine groups, while the band at 1489 334 cm −1 came from a symmetrical stretching band of -COOH. The peak at 1394 cm −1 and 882 335 cm −1 indicated the bending band of N-H and C-N from the amine group, respectively [41]. 336 In the FTIR spectra of W18O49 and NayWO3 (1:8)  27.5% and 9.7% at the third and the fourth cycle, respectively. A similar trend of remov efficiency was found in the UV light studies; however, it was higher than those obtain in the dark. For instance, the removal efficiency dropped from 94.7% to 92.5%, 40.9% an 12.9% at the second, third and fourth cycle, respectively. The improved efficiency by 0.8 14.6%, 13.4% and 3.2% for first, second, third and fourth cycle, respectively obtained aft the introduction of UV irradiation could be attributed to the adsorption process initiat by the UV light. In order to analyze the adsorption kinetics of MB on W18O49 and NayWO3, the pseud first-order kinetic and pseudo-second-order kinetic models were tested. The pseudo-fir order kinetic and pseudo-second-order kinetic models are given in Equations 3 and [44,45].  In order to analyze the adsorption kinetics of MB on W 18 O 49 and Na y WO 3 , the pseudofirst-order kinetic and pseudo-second-order kinetic models were tested. The pseudo-first-order kinetic and pseudo-second-order kinetic models are given in Equations (3) and (4) [44,45]. log(q e − q t ) = −1K 1 + logq e (3) t/q t = 1/(K 2 × q e 2 ) + t/q e (4) where q e and q t (mg/g) refer to the adsorption capacity at equilibrium and time t, respectively. K 1 and K 2 is the rate constant of pseudo-first-order equation (min −1 ) and pseudo-second-order equation (min −1 ), respectively. Figure 10a showed the non-linear relationship between log(q e − q t ) vs. t, However the adsorption processes of MB caused by the particles were pseudo-second-order kinetic isotherms shown as linear lines, as can be seen in Figure 10b. Table 5 summarizes the rate constants of the samples in MB removal. It was found that Na y WO 3 (1:16) bronze recorded the largest rate constant, i.e., 0.056 min −1 , and its MB removal rate was approximately four times faster than that of W 18 O 49 , which had a rate constant of 0.013 min −1 . In short, the Na y WO 3 bronzes did not improve the removal efficiency for MB, but they enhanced the removal rate instead. The fast kinetics of those doping sample adsorption could relate to the coulombic types of interaction between cationic dye MB and tungsten oxide [46]. The oxides have negative surface charge as confirmed by zeta-potential (as shown in Figure 8), which is due to the presence of -OH groups on the surface and oxygen vacancy in the near surface of materials. An increase of V o can be obtained via Na doping into the oxide structure; as a result, the absorption is four times faster than that of the pure W 18

Conclusions
Blue tungsten oxide (W18O49) and tungsten oxide bronze structures (NayWO3) were successfully synthesized via the simple solvothermal technique. The relationship between oxygen vacancies (Vo) and optical properties was investigated and reported. Na dopants were observed to change the band energy values and lattice parameters of the bronze framework. XPS assessments established the presence of Vo in the framework of NayWO3 bronzes. The NayWO3 samples had demonstrated a high removal efficiency of over 90% for photocatalytic MB dye adsorption under different conditions (dark, visible and UV light). Among different bronzes, the NayWO3 sample prepared using 1:8 (Na:W) atomic ratio exhibited the best MB removal performance, and its adsorption ability was attributed to the highest amount of Vo inside the crystalline framework (as evidenced by the XPS result). Vo was a source of free and localized electrons and could play an important role

Conclusions
Blue tungsten oxide (W 18 O 49 ) and tungsten oxide bronze structures (Na y WO 3 ) were successfully synthesized via the simple solvothermal technique. The relationship between oxygen vacancies (V o ) and optical properties was investigated and reported. Na dopants were observed to change the band energy values and lattice parameters of the bronze framework. XPS assessments established the presence of V o in the framework of Na y WO 3 bronzes. The Na y WO 3 samples had demonstrated a high removal efficiency of over 90% for photocatalytic MB dye adsorption under different conditions (dark, visible and UV light). Among different bronzes, the Na y WO 3 sample prepared using 1:8 (Na:W) atomic ratio exhibited the best MB removal performance, and its adsorption ability was attributed to the highest amount of V o inside the crystalline framework (as evidenced by the XPS result). V o was a source of free and localized electrons and could play an important role in the absorption mechanism for both the blue tungsten oxide and tungsten bronze structures. We had demonstrated that Na doping could improve the kinetic rate of MB removal to be approximately four times faster than that of plain W 18 O 49 nanowires.