Application of Reverse Micelle Sol–Gel Synthesis for Bulk Doping and Heteroatoms Surface Enrichment in Mo-Doped TiO2 Nanoparticles

TiO2 nanoparticles containing 0.0, 1.0, 5.0, and 10.0 wt.% Mo were prepared by a reverse micelle template assisted sol–gel method allowing the dispersion of Mo atoms in the TiO2 matrix. Their textural and surface properties were characterized by means of X-ray powder diffraction, micro-Raman spectroscopy, N2 adsorption/desorption isotherms at −196 °C, energy dispersive X-ray analysis coupled to field emission scanning electron microscopy, X-ray photoelectron spectroscopy, diffuse reflectance UV–Vis spectroscopy, and ζ-potential measurement. The photocatalytic degradation of Rhodamine B (under visible light and low irradiance) in water was used as a test reaction as well. The ensemble of the obtained experimental results was analyzed in order to discover the actual state of Mo in the final materials, showing the occurrence of both bulk doping and Mo surface species, with progressive segregation of MoOx species occurring only at a higher Mo content.


Introduction
Titanium dioxide (TiO 2 ), in virtue of its chemical stability and low toxicity, is currently used in a wide range of applications, from dye-sensitized solar cells, to sensor devices and paints [1][2][3]; most of all, TiO 2 is one of the most investigated photocatalysts (along with ZnO), as it is able to decompose several organic/inorganic pollutants in both liquid and gas phases [4][5][6].
TiO 2 may occur as anatase, rutile, or brookite phases; amorphous TiO 2 may be obtained, as well, and by properly tailoring the synthesis conditions, pure or mixed phases are obtained. While in bulk TiO 2 , rutile is the stable phase above 600 • C, synthesis methods starting from solutions containing a Ti precursor generally lead to the formation of anatase. When TiO 2 nanoparticles (NPs) are obtained, polymorphs that are usually metastable in the bulk may be stabilized instead. Among TiO 2 polymorphs, anatase has the lowest surface energy, and NPs of 15-30 nm dimension are easily obtained [7][8][9], Table 1. Textural and surface properties of the studied samples as determined by N 2 isotherms at −196 • C; FESEM; XRD, EDX and XPS (X-ray Photoelectron Spectroscopy) analyses.

Sample
BET SSA (m 2 g −1 ) Total Pore Volume (cm 3  X-ray Photoelectron Spectroscopy (XPS) analyses were performed with a Versa Probe II Scanning XPS Microprobe spectrometer (Physical Electronics GmbH, Ismaning, Germany). The measurements were done with a monochromatised Al Kα source (x-ray spot 100 µm), at a power of 24.8 W. Wide scans spectra were acquired in Fixed Analyzer Transmission (FAT) mode with a pass energy of 117.40. An electron gun was used for charge compensation (1.0V 20.0 µA). Data processing was performed by using the MultiPak software v. 9.8.0. 19.
Diffuse reflectance (DR) UV-Visible (UV-Vis) spectra of the powders were recorded in the 200-800 nm range by using a UV-Vis Varian Cary 5000 spectrophotometer (Varian Instruments, Mulgrave, Australia) equipped with an integration sphere. The DR spectra is reported as Kubelka-Munk function (Equation (1)) where R is the absolute reflectance of the layer, s the scattering coefficient, and k the molar absorption coefficient under the assumption of "infinite" thick layer.
ζ-potential was measured on a Zetasizer Nano ZSP apparatus (Malvern Instruments Ltd., Malvern, Worcestershire, UK) on suspensions of 5 mg sample in 10 mL distilled water. Before the measurement, the suspension was sonicated for 5 min and pH was adjusted by using either HCl or NH 4 OH. The corresponding ζ-potential was obtained from the electrophoretic mobility, according to the Smoluchowski's approximation.
To evaluate the photocatalytic response of the samples, Rhodamine B (RhB) was used as a model dye pollutant, the details of the photocatalytic evaluation and reactor being reported elsewhere [33]. In brief, a 9-Watt white fluorescent lamp (with maximum 3.00% UV portion) with very low irradiance (33 W/m 2 , to minimize dye sensitization effect) was used as visible light source (400-700 nm) to irradiate the dye solution containing 40 mg of synthesized sample and 50 mL of RhB dye at a concentration of 5 ppm.

Physico-Chemical Characterization of the Prepared Samples
The sol-gel chemistry of metal alkoxides is to some extent ruled by the rapid hydrolysis of the alkoxy groups, which explains why careful handling and storage of such reagents are required [34,35]. The vigorous reaction of Ti alkoxides with water leads to (undesired) formation of Ti oxo/hydroxo precipitates: the reverse micelles method is useful to control the rate of hydrolysis and condensation by the frequency of intermicellar exchanges and by the water content in the micelle core, in addition to the structure-directing role of the surfactant. The surfactant indeed provides a cage-like environment that limits the nucleation, growth, and agglomeration of the particles, which dimensions depend on the water to surfactant molar ratio and surfactant composition and morphology.
As reported by Wang et al. [36] and Chandra et al. [30] the size of reverse-micelle core depends on the synthesis parameters, predominantly the water to surfactant ratio and the nature of the solvent medium. Nevertheless, both size and shape of the NPs can be tuned by changing the polarity of hydrocarbon chain and the number polyoxyethylene groups. For instance, the non-ionic surfactant Brij-58 (linear formula C 16 H 33 (CH 2 CH 2 O) 20 OH; molar mass~1124 g/mol) allows obtaining TiO 2 NPs with an average size of about 23 nm [36]. In this study, instead, another surfactant was used, namely the Brij-98/O20 (linear formula C 18 H 35 (OCH 2 CH 2 ) 20 OH; molar mass~1150 g/mol): our purpose was to investigate the effect of another moiety on the size of the final NPs and on the overall Mo dispersion. Figure 1a reports the powder XRD patterns of the synthesized samples in the 20-90 2θ range: in agreement with the calcination temperature, the TiO 2 sample showed the main peaks of anatase (labelled by asterisks in the Figure, at the 2θ values of 25.2 (011), 37.8 (004), 47.9 (020), 53.8 (015), 54.9 (121), 62.6 (024), 68.7 (116), 70.1 (220), 74.9 (125), and 82.5 (224)), along with an additional broad and weak peak centered at ca. 30.7 2θ (circle) that can be ascribed to the (121) diffraction of brookite. The formation of such a phase, likely occurring in small amounts, can be assigned to the adopted synthesis. The XRD patterns of the Mo_1 sample did not differ much from those of TiO 2 , whereas with both Mo_5 and Mo_10 samples, two additional peaks (cross) at 27.2 and 54.4 2θ values are respectively assigned to the (110) and (211) diffraction peaks of rutile. Since the ionic radius of Mo 6+ ion (0.059 nm) is very close to that of Ti 4+ ion (0.0605 nm), Mo doping in TiO 2 mostly occurs by substitution, with formation of impurities/defects. It is generally acknowledged that the structure of rutile is more tolerant to defects than that of anatase and so, based on the current XRD results, it can be inferred that the formation of rutile was favored at higher Mo content [37,38]. The Williamson-Hall method was used to calculate the crystallite sizes reported in  [36]. In this study, instead, another surfactant was used, namely the Brij-98/O20 (linear formula C18H35(OCH2CH2)20OH; molar mass ~1150 g/mol): our purpose was to investigate the effect of another moiety on the size of the final NPs and on the overall Mo dispersion. Figure 1a reports the powder XRD patterns of the synthesized samples in the 20-90 2θ range: in agreement with the calcination temperature, the TiO2 sample showed the main peaks of anatase (labelled by asterisks in the Figure, at the 2θ values of 25.2 (011), 37.8 (004), 47.9 (020), 53.8 (015), 54.9 (121), 62.6 (024), 68.7 (116), 70.1 (220), 74.9 (125), and 82.5 (224)), along with an additional broad and weak peak centered at ca. 30.7 2θ (circle) that can be ascribed to the (121) diffraction of brookite. The formation of such a phase, likely occurring in small amounts, can be assigned to the adopted synthesis. The XRD patterns of the Mo_1 sample did not differ much from those of TiO2, whereas with both Mo_5 and Mo_10 samples, two additional peaks (cross) at 27.2 and 54.4 2θ values are respectively assigned to the (110) and (211) diffraction peaks of rutile. Since the ionic radius of Mo 6+ ion (0.059 nm) is very close to that of Ti 4+ ion (0.0605 nm), Mo doping in TiO2 mostly occurs by substitution, with formation of impurities/defects. It is generally acknowledged that the structure of rutile is more tolerant to defects than that of anatase and so, based on the current XRD results, it can be inferred that the formation of rutile was favored at higher Mo content [37,38]. The Williamson-Hall method was used to calculate the crystallite sizes reported in Table 1: showing a slight increase with Mo_1 and Mo_5 samples. No signals ascribable to MoOx phases were detected, even at the highest Mo content: such a result indicates that Mo-containing phases, if present, are likely very well dispersed and cannot be detected by XRD. The Raman spectra in Figure 1b basically confirm the XRD results: with all the samples, the Raman modes of anatase are observed at 147 (E g ), 199 (E g ), 399 (B 1g ), 519 (B 1g ) and 639 (E g ) cm −1 . With the Mo_5 and Mo_10 samples, the band at 639 cm −1 is slightly red-shifted (637 cm −1 ) with respect to that of bare TiO 2 : the most intense mode of rutile (A 1g ) occurs indeed at 636 cm −1 and so the observed shift could be ascribed to the simultaneous presence of both rutile and anatase, in agreement with the corresponding XRD patterns. Additional Raman signals (circles in Figure 1b) were observed at 246, 327, 362, and 448 cm −1 , their intensities increasing with the Mo content: they are assigned to the A 1g (246 cm −1 ), B 1g (327 and 448 cm −1 ), and B 2g (362 cm −1 ) modes of brookite. The most intense band of brookite (B 1g mode) was usually found at 152 cm −1 and, here, is likely superposed to that of anatase. Indeed, the maximum of the main peak was blue-shifted with the Mo_10 sample, indicating a strong interaction of the heteroatoms with the TiO 2 matrix.
With the Mo_5 and Mo_10 samples, two additional Raman signals were observed: a band occurring at 944 cm −1 with the Mo_5 sample and at 956 cm −1 with the Mo_10 sample (asterisk), and a broad signal occurring in the 760-880 cm −1 range of the Mo_10 spectrum (arrow). The former signal was assigned to Mo=O groups stretching [39,40]: with systems where a comparable amount of MoO 3 was supported on TiO 2 by incipient wetness impregnation, a similar band was observed in the 934-954 cm −1 range, shifting to higher wavenumbers with the Mo content, and having a broad and asymmetric shape, as here.  [41]. Figure 2 shows the N 2 adsorption/desorption isotherms at −196 • C on both pure and Mo-doped samples, and the corresponding pore size distribution (PSD) as calculated by applying the BJH method to the isotherm desorption branch. All the samples show Type IV isotherms with Type H2 hysteresis loop, due to N 2 condensation within inter-particles porosity. The BET SSA value (Table 1) was almost unaffected by the presence of Mo for heteroatom loadings up to 5 wt.%, while it increases with the Mo_10 sample.
The bare TiO 2 showed a sharp Pore Size Distribution (PSD) in the 3-6 nm range (see Figure 2b), suggesting a homogeneous distribution of particles size. The same sample showed also the smallest particle size (see Table 1) and a quasi-round particle shape (see Figure 3a): from a merely geometrical point of view, the inter-particle porosity is expected to be rather homogeneous with smaller dimensions with respect to the particle size. Interestingly, the PSD became broader with Mo-doped samples. Actually, both the size and shape of the particles could affect the BET SSA and the PSDs, and, probably, less homogeneous (in both size and shape) particles were obtained in the presence of Mo. In particular, the Mo_5 sample showed the broadest PSD (in the 3-14 nm range): the sample showed indeed nearly the same BET SSA of Mo_1 (Table 1), but a larger pore volume. This could be due to the occurrence of NPs characterized by heterogeneous size and shape (see Figure 3b), although some aggregation/agglomeration phenomena may have occurred in Mo_5 as well.   (Table 1) and, indeed, the Mo_5 sample showed NPs with more heterogeneous size and shape, in agreement with the sample PSD (see Figure 2b).  NPs with quasi-round shape are observed, but the addition of Mo leads to a slight size increase due to induced structural changes (Table 1) and, indeed, the Mo_5 sample showed NPs with more heterogeneous size and shape, in agreement with the sample PSD (see Figure 2b).   (Table 1) and, indeed, the Mo_5 sample showed NPs with more heterogeneous size and shape, in agreement with the sample PSD (see Figure 2b). The EDX analysis allowed us determining the Mo/Ti atomic ratios reported in Table 1, showing that in the explored composition range, the actual Mo content was very close to the nominal value. Figure 4 reports DR UV-Vis spectra of the samples. As expected, the TiO2 sample absorbed only below 400 nm, whereas introduction of Mo brought about two effects: a slight red-shift of the absorption edge and the appearance of a broad absorption centered at ca. 550 nm readily assigned to sub-band transitions related to mid-band gaps formed by Mo doping [42]. The red-shift of the absorption edge shows that Mo doping is modifying the band gap (Eg): the corresponding Eg values, as obtained by the Tauc's plot method (not shown), are indeed 3.07 eV (TiO2), 2.86 ev (Mo_1), 2.58 eV (Mo_5) and 2.69 eV (Mo_10). Interestingly, the Mo_5 sample showed the lowest Eg value: at higher contents, Mo likely tends to form surface species, rather than being dispersed in the bulk, as also confirmed by Raman and XP spectroscopies (vide supra). The ζ-potential measurements ( Figure 5) show that the bare TiO2 has a Point of Zero Charge (PZC) of 3.6, that is, a value lower than usual for TiO2: for instance, Degussa P25 (with particle size between 20 and 40 nm) has a PZC around 6.2-6.9 [43,44]; for pure TiO2 NPs (with diameter in the range of 14-33 nm), a PZC of 6.8 [45]   The EDX analysis allowed us determining the Mo/Ti atomic ratios reported in Table 1, showing that in the explored composition range, the actual Mo content was very close to the nominal value. Figure 4 reports DR UV-Vis spectra of the samples. As expected, the TiO 2 sample absorbed only below 400 nm, whereas introduction of Mo brought about two effects: a slight red-shift of the absorption edge and the appearance of a broad absorption centered at ca. 550 nm readily assigned to sub-band transitions related to mid-band gaps formed by Mo doping [42]. The red-shift of the absorption edge shows that Mo doping is modifying the band gap (E g ): the corresponding E g values, as obtained by the Tauc's plot method (not shown), are indeed 3.07 eV (TiO 2 ), 2.86 ev (Mo_1), 2.58 eV (Mo_5) and 2.69 eV (Mo_10). Interestingly, the Mo_5 sample showed the lowest E g value: at higher contents, Mo likely tends to form surface species, rather than being dispersed in the bulk, as also confirmed by Raman and XP spectroscopies (vide supra). The EDX analysis allowed us determining the Mo/Ti atomic ratios reported in Table 1, showing that in the explored composition range, the actual Mo content was very close to the nominal value. Figure 4 reports DR UV-Vis spectra of the samples. As expected, the TiO2 sample absorbed only below 400 nm, whereas introduction of Mo brought about two effects: a slight red-shift of the absorption edge and the appearance of a broad absorption centered at ca. 550 nm readily assigned to sub-band transitions related to mid-band gaps formed by Mo doping [42]. The red-shift of the absorption edge shows that Mo doping is modifying the band gap (Eg): the corresponding Eg values, as obtained by the Tauc's plot method (not shown), are indeed 3.07 eV (TiO2), 2.86 ev (Mo_1), 2.58 eV (Mo_5) and 2.69 eV (Mo_10). Interestingly, the Mo_5 sample showed the lowest Eg value: at higher contents, Mo likely tends to form surface species, rather than being dispersed in the bulk, as also confirmed by Raman and XP spectroscopies (vide supra). The ζ-potential measurements ( Figure 5) show that the bare TiO2 has a Point of Zero Charge (PZC) of 3.6, that is, a value lower than usual for TiO2: for instance, Degussa P25 (with particle size between 20 and 40 nm) has a PZC around 6.2-6.9 [43,44]; for pure TiO2 NPs (with diameter in the range of 14-33 nm), a PZC of 6.8 [45]   The ζ-potential measurements ( Figure 5) show that the bare TiO 2 has a Point of Zero Charge (PZC) of 3.6, that is, a value lower than usual for TiO 2 : for instance, Degussa P25 (with particle size between 20 and 40 nm) has a PZC around 6.2-6.9 [43,44]; for pure TiO 2 NPs (with diameter in the range of 14-33 nm), a PZC of 6.8 [45] was reported; Allard et al. found a PZC of 6.1 with commercial anatase NPs (with a diameter of ca. 20 nm) [46]; Al-Hetlani et al. found a PZC of 5.98 for smaller anatase NPs (around 7.2 nm [47]); and Huijun et al. reported a PZC of 6.2 for anatase NPs (5-10 nm) [48]. The much lower value of PZC reported here may be ascribed, rather than to the size of TiO 2 NPs, to the type of synthesis [49,50] that probably favors the formation of a very acidic surface, in agreement with previous work [43]. Addition of Mo leads to a further and progressive decrease of the PZC, even at the lowest Mo content (sample Mo_1). Such results confirm that this type of synthesis allows distribution of the heteroatoms not only in the bulk, but also at the NPs' surface, with some superficial molybdenum ions lowering the PZC, as they are strong Lewis sites.
Materials 2019, 12 FOR PEER REVIEW 9 [48]. The much lower value of PZC reported here may be ascribed, rather than to the size of TiO2 NPs, to the type of synthesis [49,50] that probably favors the formation of a very acidic surface, in agreement with previous work [43]. Addition of Mo leads to a further and progressive decrease of the PZC, even at the lowest Mo content (sample Mo_1). Such results confirm that this type of synthesis allows distribution of the heteroatoms not only in the bulk, but also at the NPs' surface, with some superficial molybdenum ions lowering the PZC, as they are strong Lewis sites. The ζ potential measurements allowed us to figure out that even at low Mo content, the surface of the NPs is affected by the presence of Mo: the ancillary XPS measurements (Supplementary Information) allowed us to measure the surface Mo/Ti atomic ratio in the studied samples: the corresponding values (reported in Table 1) show that the NPs surface is enriched in Mo atoms with respect to the bulk, in agreement with the type of synthesis adopted.
These obtained samples are negatively charged in a wide pH range ( Figure 5), and so they should preferably interact with positively charged species, such as the Rhodamine B (RhB) dye [51].
Indeed, RhB (Scheme 1) is a cationic Xanthene type dye, characterized by the presence of both diethylamine and carboxylic groups and is commonly used as a model dye pollutant [52,53]. For these reasons, the photocatalytic RhB degradation was studied in the presence of the Mo-doped samples and of the bare TiO2. Visible light irradiation with a low irradiance value of 33 W/m 2 was used to avoid or at least minimize any dye-sensitization effect and to test the feasibility of sun-driven photocatalytic reaction [33].  The ζ potential measurements allowed us to figure out that even at low Mo content, the surface of the NPs is affected by the presence of Mo: the ancillary XPS measurements (Supplementary Information) allowed us to measure the surface Mo/Ti atomic ratio in the studied samples: the corresponding values (reported in Table 1) show that the NPs surface is enriched in Mo atoms with respect to the bulk, in agreement with the type of synthesis adopted.
These obtained samples are negatively charged in a wide pH range ( Figure 5), and so they should preferably interact with positively charged species, such as the Rhodamine B (RhB) dye [51].
Indeed, RhB (Scheme 1) is a cationic Xanthene type dye, characterized by the presence of both diethylamine and carboxylic groups and is commonly used as a model dye pollutant [52,53]. For these reasons, the photocatalytic RhB degradation was studied in the presence of the Mo-doped samples and of the bare TiO 2 . Visible light irradiation with a low irradiance value of 33 W/m 2 was used to avoid or at least minimize any dye-sensitization effect and to test the feasibility of sun-driven photocatalytic reaction [33].
Materials 2019, 12 FOR PEER REVIEW 9 [48]. The much lower value of PZC reported here may be ascribed, rather than to the size of TiO2 NPs, to the type of synthesis [49,50] that probably favors the formation of a very acidic surface, in agreement with previous work [43]. Addition of Mo leads to a further and progressive decrease of the PZC, even at the lowest Mo content (sample Mo_1). Such results confirm that this type of synthesis allows distribution of the heteroatoms not only in the bulk, but also at the NPs' surface, with some superficial molybdenum ions lowering the PZC, as they are strong Lewis sites. The ζ potential measurements allowed us to figure out that even at low Mo content, the surface of the NPs is affected by the presence of Mo: the ancillary XPS measurements (Supplementary Information) allowed us to measure the surface Mo/Ti atomic ratio in the studied samples: the corresponding values (reported in Table 1) show that the NPs surface is enriched in Mo atoms with respect to the bulk, in agreement with the type of synthesis adopted.
These obtained samples are negatively charged in a wide pH range ( Figure 5), and so they should preferably interact with positively charged species, such as the Rhodamine B (RhB) dye [51].
Indeed, RhB (Scheme 1) is a cationic Xanthene type dye, characterized by the presence of both diethylamine and carboxylic groups and is commonly used as a model dye pollutant [52,53]. For these reasons, the photocatalytic RhB degradation was studied in the presence of the Mo-doped samples and of the bare TiO2. Visible light irradiation with a low irradiance value of 33 W/m 2 was used to avoid or at least minimize any dye-sensitization effect and to test the feasibility of sun-driven photocatalytic reaction [33].  The obtained RhB degradation curves are shown in Figure 6, where, for comparison, the results concerning two blank experiments carried out without any catalyst (mere photolysis) and in the presence of a commercial TiO 2 sample (Degussa P25) are reported, showing, respectively, that under Vis light, RhD photolysis occurs to a minor extent and that the commercial TiO 2 was the least active of the tested samples. Concerning the photocatalytic activity of the bare TiO 2 (E g = 3.07 eV), XRD patterns shows that the sample also contained some brookite (see Figure 1): other mixed TiO 2 phases containing brookite showed, indeed, promising photodegradation activity towards RhB under Vis light [54]. As compared to the bare TiO 2 , the Mo_10 sample showed similar behavior during the dark step, whereas under illumination, its RhB degradation efficiency was even lower than that of the bare TiO 2 at a longer reaction time. Notwithstanding its smaller band gap (E g = 2.69 eV) and lower PZC, which should favor, respectively, photocatalytic activity and the interaction with diethylamine groups (protonated in the adopted reaction conditions), the (high) Mo content of the Mo_10 sample likely had a detrimental effect on RhB degradation. At a higher content, Mo tends to form surface species, as shown by both Raman and XPS spectroscopies (the Mo_10 sample has indeed a slightly larger band gap (E g = 2.69 eV) than Mo_5 (E g = 2.58 eV)). Nonetheless, the formation of surface polymolibdates (as detected by Raman spectroscopy) could enhance surface electron/hole recombination, finally decreasing photocatalytic activity. Interestingly, an improved kinetic rate was achieved with the Mo_5 sample, characterized by the smallest E g value (E g = 2.58 eV). Such results suggest that Mo_5 has an optimum dopant concentration allowing a better exploitation of the light (due to the lowest band gap), notwithstanding the adsorption properties of the sample in dark conditions were worse than both TiO 2 and Mo_10, likely due to the larger particles size of Mo_5 (22 nm, see Table 1). Similar considerations can be drawn for the adsorption behavior of the Mo_1 sample, where the lower Mo content was instead responsible of its lower activity with respect to Mo_5. Concerning the kind of dye-surface interaction, complex phenomena may occur, besides electrostatic attraction between the protonated diethylamine groups of the dye and the surface negative charge of the samples, as RhB also contains carboxylate groups and some positive surface charges may occur, especially when Mo ions are at the surface of the NPs. However, the interaction between protonated diethylamine groups of RhB and the negatively charged surface of the prepared samples could improve the degradation rate by cleavage of the Xantene group. This result could be further confirmed from the shift of the main RhB peak from 554 nm to 535 nm observed in the dye UV-Vis spectrum (inset to Figure 6): such shifts mainly occur when the dye undergoes photocatalytic degradation through a de-ethylation process/route and formation of triethylrhodamine, diethylrhodamine, and ethylrhodamine, with a different λ max , at 555, 539, 522 nm, repectively [52,53].
The obtained RhB degradation curves are shown in Figure 6, where, for comparison, the results concerning two blank experiments carried out without any catalyst (mere photolysis) and in the presence of a commercial TiO2 sample (Degussa P25) are reported, showing, respectively, that under Vis light, RhD photolysis occurs to a minor extent and that the commercial TiO2 was the least active of the tested samples. Concerning the photocatalytic activity of the bare TiO2 (Eg = 3.07 eV), XRD patterns shows that the sample also contained some brookite (see Figure 1): other mixed TiO2 phases containing brookite showed, indeed, promising photodegradation activity towards RhB under Vis light [54]. As compared to the bare TiO2, the Mo_10 sample showed similar behavior during the dark step, whereas under illumination, its RhB degradation efficiency was even lower than that of the bare TiO2 at a longer reaction time. Notwithstanding its smaller band gap (Eg = 2.69 eV) and lower PZC, which should favor, respectively, photocatalytic activity and the interaction with diethylamine groups (protonated in the adopted reaction conditions), the (high) Mo content of the Mo_10 sample likely had a detrimental effect on RhB degradation. At a higher content, Mo tends to form surface species, as shown by both Raman and XPS spectroscopies (the Mo_10 sample has indeed a slightly larger band gap (Eg = 2.69 eV) than Mo_5 (Eg = 2.58 eV)). Nonetheless, the formation of surface polymolibdates (as detected by Raman spectroscopy) could enhance surface electron/hole recombination, finally decreasing photocatalytic activity. Interestingly, an improved kinetic rate was achieved with the Mo_5 sample, characterized by the smallest Eg value (Eg = 2.58 eV). Such results suggest that Mo_5 has an optimum dopant concentration allowing a better exploitation of the light (due to the lowest band gap), notwithstanding the adsorption properties of the sample in dark conditions were worse than both TiO2 and Mo_10, likely due to the larger particles size of Mo_5 (22 nm, see Table 1). Similar considerations can be drawn for the adsorption behavior of the Mo_1 sample, where the lower Mo content was instead responsible of its lower activity with respect to Mo_5. Concerning the kind of dye-surface interaction, complex phenomena may occur, besides electrostatic attraction between the protonated diethylamine groups of the dye and the surface negative charge of the samples, as RhB also contains carboxylate groups and some positive surface charges may occur, especially when Mo ions are at the surface of the NPs. However, the interaction between protonated diethylamine groups of RhB and the negatively charged surface of the prepared samples could improve the degradation rate by cleavage of the Xantene group. This result could be further confirmed from the shift of the main RhB peak from 554 nm to 535 nm observed in the dye UV-Vis spectrum (inset to Figure 6): such shifts mainly occur when the dye undergoes photocatalytic degradation through a de-ethylation process/route and formation of triethylrhodamine, diethylrhodamine, and ethylrhodamine, with a different λmax, at 555, 539, 522 nm, repectively [52,53].

Conclusions
Both pure and Mo-doped TiO 2 NPs (with a size of ca. 10-20 nm) were obtained by a reverse micelle sol-gel synthesis method, allowing the dispersion of Mo both in the bulk and at the surface of the NPs, and substantially avoiding segregation of crystalline MoO 3 event with 10 wt.% Mo.
The 5 wt.% Mo content was found to provide an optimal lowering of the band gap (from 3.07 V for the bare TiO 2 to 2.58 eV), which resulted in the fastest kinetics during the photocatalytic degradation of the model dye Rhodamine B (used here as a characterization technique).
The synthesis also led to a very acidic (polar) surface, even in the absence of Mo: the resulting NPs were indeed negatively charged in a wide pH range. This surface negative charge, however, did not enhance the degradation of the dye studied here, especially at the highest Mo content, because the surface polymolibdate species likely acted as recombination centers of electron/hole pair. Despite all this, the high acidity of the NPs surface could be exploited in applications requiring a very polar surface, due to the possibility to polarize some organic pollutant, finally promoting its adsorption and consequent degradation.