Enhanced Performance of Electrospun Nanoﬁbrous TiO 2 / g-C 3 N 4 Photocatalyst in Photocatalytic Degradation of Methylene Blue

: TiO 2 / g-C 3 N 4 (GNT) ﬁbers with 1%, 2.5% and 5% (wt%) ratios have been synthesized via one-step electrospinning using polyvinylpyrrolidone (PVP) polymer. Results showed mesoporous ﬁbrous catalysts consisted of anatase (80.0–85.1%) and rutile phase (14.9–20.0%), with diameter between 200–300 nm and band gap lower than 3.0 eV conﬁrming the absorption shift to visible-light region. The formation of • OH radicals and methylene blue dye degradation increases as the g-C 3 N 4 doping percent also increases, following the trend TiO 2 < GNT1% ≈ GNT2.5% < GNT5%. A z-scheme mechanism is attributed to the photocatalytic performance conﬁrming the potential for green chemistry and environmental technology applications. The thermal behavior of all electrospun PVP/TTIP/g-C 3 N 4 mixtures is The decomposition of fibers takes place in three not well distinguished steps. The first endothermic step takes place up to a temperature of about 150 °C and is attributed to the


Introduction
Heterogeneous photocatalysis is among the advanced oxidation processes (AOP's) with growing applications in water and air treatment for the removal of various pollutants and disinfection. TiO 2 is the most used photocatalyst because of its high chemical and photochemical stability, high photocatalytic activity and low cost. However, the photocatalytic activity of TiO 2 is constrained by the relatively high recombination rate of electron-hole (e − -h + ) pairs and the low efficiency for solar light due to the band gap of 3.0-3.2 eV. Therefore, much effort has been devoted to design heterostructured photocatalysts of TiO 2 with other semiconductors with narrower band gap and suitable conduction and valence band potential levels. The photocatalytic properties of composite materials/heterostructures presented significant advantages like the decrease in electron-hole (e − -h + ) pair recombination rate, the absorption extension to wavelengths in the visible light region and the increase of the reactant adsorption into the catalyst surface [1][2][3].
Graphitic carbon nitride (g-C 3 N 4 ) is a promising organic semiconductor with the focus of recent studies on photocatalytic applications. It has relatively narrow band gap (2.7 eV) with two-dimensional (2D) nanostructure, high thermal (up to 600 • C in air) and chemical stability due to s-triazine ring structure, high resistance in acidic and basic solutions and insolubility in common solvents as ethanol and water [4][5][6]. Moreover, is non-toxic and can be synthesized easily with low cost precursor compounds like cyanamide, urea and melamine. The g-C 3 N 4 exhibits also the potential for degrading organic pollutants under visible light irradiation. However, its photocatalytic efficiency is quite limited due to the fast recombination of photogenerated electron-holes pairs [7].
Presently, there are many works exploring the TiO 2 /g-C 3 N 4 heterojunction structure via various methods such as in situ growth TiO 2 on g-C 3 N 4 sheets [8], single-pot micro emulsion [9],

XRD Patterns and FT-IR Spectra
The XRD patterns for g-C 3 N 4 , TiO 2 and g-C 3 N 4 /TiO 2 composite fibers are shown in Figure S1 and Figure 1, respectively. For fibrous TiO 2 and g-C 3 N 4 /TiO 2 composites all the diffraction peaks are well matched to anatase TiO 2 crystal phase using the Joint Committee on Powder Diffraction Standards (JCPDS no. 86-1157). Low amounts (<20%) of rutile (JCPDS no. 86-1157) phase were also observed. For g-C 3 N 4 (JCPDS no. 87-1522) two characteristic intense peaks at 13.1 • and 27.4 • were observed. The peak at 2θ = 13.1 • corresponds to the (100) plane of g-C 3 N 4 and it is attributed to the in-plane repetitive and continuous heptazine network. The peak located at 2θ = 27.4 • corresponds to the (002) plane of g-C 3 N 4 and it is assigned to the stacking of the conjugated aromatic system [8,[11][12][13]. Table 1 summarizes the percentages of anatase and rutile crystal phases and the crystal sizes of the prepared samples determined by the Scherrer equation [20]. The patterns (2θ • angles and peak shape) for g-C 3 N 4 /TiO 2 did not significantly changed as regards to the reference TiO 2 indicating that the coupling with g-C 3 N 4 did not alter considerably the lattice structure of TiO 2 . The loading of g-C 3 N 4 had no effect on the catalyst crystal size and the ratio of anatase to rutile crystal phases.  The FT-IR spectra of electrospun fibers of TiO2 and TiO2/g-C3N4 composite photocatalysts are presented in Figure 2. For all the fibers, the peaks at 680 and 450 cm −1 correspond to the Ti-O-Ti stretching vibration. Also, the peak at 3434 cm −1 corresponds to the stretching vibration mode O-H bonds of free water molecules and Ti-OH vibrations. Additionally, Figures S2 and S3 present the FT-IR spectra of PVP fibers and g-C3N4, respectively. The peaks of g-C3N4 can be observed at 1573, 1465, 1403, 1319 and 1241 cm −1 confirming the stretching vibration of C-N (-C)-C or C-NH-C heterocycles. The peaks between 3300 and 3000 cm −1 are related to the stretching vibration of N-H and the peak at 1637 cm −1 correspond to C=N bending vibration. The peak at 810 cm −1 is the most characteristic for g-C3N4 which is attributed to the out-of-plane bending vibration characteristic of heptazine rings [15,21]. Also, the characteristic peak of PVP fibers is the peak at 1657 cm −1 , which confirms the presence of C=O bonds. The peaks at 3434 cm −1 and 2955 cm −1 correspond to O-H vibration and C-H asymmetric stretching vibration, respectively. In conclusion, the FT-IR spectra of the prepared catalysts presented the main characteristic peaks of both TiO2 and g-C3N4 suggesting the formation of composite heterostructures while some traces of PVP chemical functions (C=O bonds) still appeared suggesting the concurrent carbonaceous modification of TiO2.  The FT-IR spectra of electrospun fibers of TiO 2 and TiO 2 /g-C 3 N 4 composite photocatalysts are presented in Figure 2. For all the fibers, the peaks at 680 and 450 cm −1 correspond to the Ti-O-Ti stretching vibration. Also, the peak at 3434 cm −1 corresponds to the stretching vibration mode O-H bonds of free water molecules and Ti-OH vibrations. Additionally, Figures S2 and S3 present the FT-IR spectra of PVP fibers and g-C 3 N 4 , respectively. The peaks of g-C 3 N 4 can be observed at 1573, 1465, 1403, 1319 and 1241 cm −1 confirming the stretching vibration of C-N (-C)-C or C-NH-C heterocycles. The peaks between 3300 and 3000 cm −1 are related to the stretching vibration of N-H and the peak at 1637 cm −1 correspond to C=N bending vibration. The peak at 810 cm −1 is the most characteristic for g-C 3 N 4 which is attributed to the out-of-plane bending vibration characteristic of heptazine rings [15,21]. Also, the characteristic peak of PVP fibers is the peak at 1657 cm −1 , which confirms the presence of C=O bonds. The peaks at 3434 cm −1 and 2955 cm −1 correspond to O-H vibration and C-H asymmetric stretching vibration, respectively. In conclusion, the FT-IR spectra of the prepared catalysts presented the main characteristic peaks of both TiO 2 and g-C 3 N 4 suggesting the formation of composite heterostructures while some traces of PVP chemical functions (C=O bonds) still appeared suggesting the concurrent carbonaceous modification of TiO 2 . The FT-IR spectra of electrospun fibers of TiO2 and TiO2/g-C3N4 composite photocatalysts are presented in Figure 2. For all the fibers, the peaks at 680 and 450 cm −1 correspond to the Ti-O-Ti stretching vibration. Also, the peak at 3434 cm −1 corresponds to the stretching vibration mode O-H bonds of free water molecules and Ti-OH vibrations. Additionally, Figures S2 and S3 present the FT-IR spectra of PVP fibers and g-C3N4, respectively. The peaks of g-C3N4 can be observed at 1573, 1465, 1403, 1319 and 1241 cm −1 confirming the stretching vibration of C-N (-C)-C or C-NH-C heterocycles. The peaks between 3300 and 3000 cm −1 are related to the stretching vibration of N-H and the peak at 1637 cm −1 correspond to C=N bending vibration. The peak at 810 cm −1 is the most characteristic for g-C3N4 which is attributed to the out-of-plane bending vibration characteristic of heptazine rings [15,21]. Also, the characteristic peak of PVP fibers is the peak at 1657 cm −1 , which confirms the presence of C=O bonds. The peaks at 3434 cm −1 and 2955 cm −1 correspond to O-H vibration and C-H asymmetric stretching vibration, respectively. In conclusion, the FT-IR spectra of the prepared catalysts presented the main characteristic peaks of both TiO2 and g-C3N4 suggesting the formation of composite heterostructures while some traces of PVP chemical functions (C=O bonds) still appeared suggesting the concurrent carbonaceous modification of TiO2.

Morphology and Thermal Analysis
The nitrogen adsorption-desorption isotherms for the studied catalysts and g-C 3 N 4 are displayed in Figure 3 and Figure S4, respectively. The synthesized fibers are typical mesoporous materials with type IVa adsorption isotherms with an H3 hysteresis loop type according to the IUPAC classification. BET specific surface area for g-C 3 N 4 /TiO 2 composites ranged between 40 m 2 g −1 and 47 m 2 g −1 . The pure g-C 3 N 4 presented a specific surface of 35.2 m 2 g −1 and a narrow hysteresis that did not present the characteristics of mesoporous structure. The calculated surface area, pore diameter and total pore volume of the prepared catalysts are shown in Table 1. The nitrogen adsorption-desorption isotherms for the studied catalysts and g-C3N4 are displayed in Figures 3 and S4, respectively. The synthesized fibers are typical mesoporous materials with type IVa adsorption isotherms with an H3 hysteresis loop type according to the IUPAC classification. BET specific surface area for g-C3N4/TiO2 composites ranged between 40 m 2 g −1 and 47 m 2 g −1 . The pure g-C3N4 presented a specific surface of 35.2 m 2 g −1 and a narrow hysteresis that did not present the characteristics of mesoporous structure. The calculated surface area, pore diameter and total pore volume of the prepared catalysts are shown in Table 1. The structural features of all catalysts were also studied by SEM. The as spun fibers and the fibers after calcinations appeared in Figures S5 and 4, respectively. The diameter of as spun fibers varied between 300-350 nm and between 200-350 nm for the calcinated ones. Moreover, in Figure S6 it is shown the structure of bulk g-C3N4 consisting of stacking sheets. The structural features of all catalysts were also studied by SEM. The as spun fibers and the fibers after calcinations appeared in Figure S5 and Figure 4, respectively. The diameter of as spun fibers varied between 300-350 nm and between 200-350 nm for the calcinated ones. Moreover, in Figure S6 it is shown the structure of bulk g-C 3 N 4 consisting of stacking sheets.  In order to study the thermal degradation of the prepared electrospun precursors the thermal analyses were performed and presented in Figure 5 for thermal gravimetric analysis (TGA) curves and Figure S7 for differential scanning calorimetry (DSC) curves. Thermal analysis pattern of electrospun PVP shows degradation in three steps. The first endothermic step takes place up to temperature of 110 °C with a mass loss of about 13.3% and is attributed to the removal of remained solvent and adsorbed water. The second step is accompanied by two exothermic and one endothermic stage which could be attributed to the main decomposition of PVP. This step takes place at temperature range from about 300 to 475 °C, with a weight loss of about 67.8. The final step is attributed to combustion of the yielding organic matter as is denoted in DSC curve from one very broad exothermic peak up to temperature of about 665 °C. An exothermic peak at 197 °C, without mass loss, is takes place after the theoretical glass transition of PVP. With increase in molecular weight, PVP Tg increases to a limiting value of 180 °C [22]. The Tg response does not appear on the DSC curve due to the high moisture content, in both the electrospun sample and the commercial sample (the latter does not appear). Elishav et al. [23] studied the thermal shrinkage of electrospun PVP fibers and found that the shrinkage has a marked knee at 150-180 °C and this range coincides with the polymer glass transition temperature. They suggested that, as a result of a high rate fibers' drying due to the intense solvents evaporation, an internal stress is applied which can relax during thermal treatment giving rise to non-uniform shrinkage that leads to shape loss.
The exothermic peak at 197 °C can be attributed to shrinkage phenomena. The fact that this exothermic peak does not appear in the commercial sample confirms this hypothesis.
The thermal behavior of all electrospun PVP/TTIP/g-C3N4 mixtures is similar. The decomposition of fibers takes place in three not well distinguished steps. The first endothermic step takes place up to a temperature of about 150 °C and is attributed to the removal of remained In order to study the thermal degradation of the prepared electrospun precursors the thermal analyses were performed and presented in Figure 5 for thermal gravimetric analysis (TGA) curves and Figure S7 for differential scanning calorimetry (DSC) curves. Thermal analysis pattern of electrospun PVP shows degradation in three steps. The first endothermic step takes place up to temperature of 110 • C with a mass loss of about 13.3% and is attributed to the removal of remained solvent and adsorbed water. The second step is accompanied by two exothermic and one endothermic stage which could be attributed to the main decomposition of PVP. This step takes place at temperature range from about 300 to 475 • C, with a weight loss of about 67.8. The final step is attributed to combustion of the yielding organic matter as is denoted in DSC curve from one very broad exothermic peak up to temperature of about 665 • C. An exothermic peak at 197 • C, without mass loss, is takes place after the theoretical glass transition of PVP. With increase in molecular weight, PVP T g increases to a limiting value of 180 • C [22]. The T g response does not appear on the DSC curve due to the high moisture content, in both the electrospun sample and the commercial sample (the latter does not appear). Elishav et al. [23] studied the thermal shrinkage of electrospun PVP fibers and found that the shrinkage has a marked knee at 150-180 • C and this range coincides with the polymer glass transition temperature. They suggested that, as a result of a high rate fibers' drying due to the intense solvents evaporation, an internal stress is applied which can relax during thermal treatment giving rise to non-uniform shrinkage that leads to shape loss.
The exothermic peak at 197 • C can be attributed to shrinkage phenomena. The fact that this exothermic peak does not appear in the commercial sample confirms this hypothesis.
The thermal behavior of all electrospun PVP/TTIP/g-C 3 N 4 mixtures is similar. The decomposition of fibers takes place in three not well distinguished steps. The first endothermic step takes place up to a temperature of about 150 • C and is attributed to the removal of remained solvents. The other two steps were exothermic and are attributed to the degradation of PVP as well as to dehydration of titania gel. The exothermic crystallization peak of PVP at 197 • C is not observed which may suggest that significant covalent bonds existed between the titania gel and the polymer matrix, and therefore the shrinking process is not observed. On the other hand, the PVP degradation end is observed at about 560 • C, that is, 100 degree lower, indicating that titania catalyzes it. solvents. The other two steps were exothermic and are attributed to the degradation of PVP as well as to dehydration of titania gel. The exothermic crystallization peak of PVP at 197 °C is not observed which may suggest that significant covalent bonds existed between the titania gel and the polymer matrix, and therefore the shrinking process is not observed. On the other hand, the PVP degradation end is observed at about 560 °C, that is, 100 degree lower, indicating that titania catalyzes it. The measured DLS median hydrodynamic particle size diameters of TiO2 fibers, GNT1%, GNT2.5% and GNT5% in aqueous suspensions after sonication for 10 min were 0.321, 0.239, 0.242 and 0.325 μm, respectively.

UV-Vis Diffuse Reflectance and Fluorescence Spectra
The optical properties of prepared composite catalysts were investigated by the UV-Vis diffuse reflection spectroscopy. Figure 6 displays the UV-Vis absorption spectra of fibrous materials and Figure 7 the determined energy band gaps (Eg) of each photocatalyst using the Kübelka-Münk function. The absorption edges of TiO2 fibers, GNT1%, GNT2.5% and GNT5% were determined at 442, 424, 444 and 426 nm, while the corresponding energy band gaps of TiO2 fibers, GNT1%, GNT2.5% and GNT5% were 2.8, 2.92, 2.8 and 2.91 eV, respectively. The presence of the red-shift in the reference TiO2 fibers could be attributed to residual carbon modification as revealed by FT-IR spectroscopy. The results indicate the improved response to visible light for all prepared fibrous catalysts. The measured DLS median hydrodynamic particle size diameters of TiO 2 fibers, GNT1%, GNT2.5% and GNT5% in aqueous suspensions after sonication for 10 min were 0.321, 0.239, 0.242 and 0.325 µm, respectively.

UV-Vis Diffuse Reflectance and Fluorescence Spectra
The optical properties of prepared composite catalysts were investigated by the UV-Vis diffuse reflection spectroscopy. Figure 6 displays the UV-Vis absorption spectra of fibrous materials and Figure 7 the determined energy band gaps (E g ) of each photocatalyst using the Kübelka-Münk function. The absorption edges of TiO 2 fibers, GNT1%, GNT2.5% and GNT5% were determined at 442, 424, 444 and 426 nm, while the corresponding energy band gaps of TiO 2 fibers, GNT1%, GNT2.5% and GNT5% were 2.8, 2.92, 2.8 and 2.91 eV, respectively. The presence of the red-shift in the reference TiO 2 fibers could be attributed to residual carbon modification as revealed by FT-IR spectroscopy. The results indicate the improved response to visible light for all prepared fibrous catalysts.  The photocatalytic ability of composite catalysts to generate • OH radicals was determined by fluorescence spectroscopy. The experimental results (Figures 8, 9, S8 and Table 2) showed that the kinetics of • OH radicals formation followed the trend k GNT 5% > k GNT 2.5% ≈ k GNT 1% > k TiO2 fibers which is consistent with the trend of the fibrous kinetics for the degradation of methylene blue. The photogeneration rate of • OH radicals for TiO2 fibers, GNT 1%, GNT 2.5% and GNT 5% were 0.288, 0.297, 0.298 and 0.348 μM min −1 ., respectively. As a result, the increase of g-C3N4 content in the composite fibers led to higher concentrations of generated • OH radicals. The photocatalytic ability of composite catalysts to generate • OH radicals was determined by fluorescence spectroscopy. The experimental results ( Figure 8, Figure 9, Figure S8 and Table 2) showed that the kinetics of • OH radicals formation followed the trend k GNT 5% > k GNT 2.5% ≈ k GNT 1% > k TiO2 fibers which is consistent with the trend of the fibrous kinetics for the degradation of methylene blue. The photogeneration rate of • OH radicals for TiO 2 fibers, GNT 1%, GNT 2.5% and GNT 5% were 0.288, 0.297, 0.298 and 0.348 µM min −1 ., respectively. As a result, the increase of g-C 3 N 4 content in the composite fibers led to higher concentrations of generated • OH radicals. The photocatalytic ability of composite catalysts to generate • OH radicals was determined by fluorescence spectroscopy. The experimental results (Figures 8, 9, S8 and Table 2) showed that the kinetics of • OH radicals formation followed the trend k GNT 5% > k GNT 2.5% ≈ k GNT 1% > k TiO2 fibers which is consistent with the trend of the fibrous kinetics for the degradation of methylene blue. The photogeneration rate of • OH radicals for TiO2 fibers, GNT 1%, GNT 2.5% and GNT 5% were 0.288, 0.297, 0.298 and 0.348 μM min −1 ., respectively. As a result, the increase of g-C3N4 content in the composite fibers led to higher concentrations of generated • OH radicals.

. Photocatalytic Activity and Recyclability
The photocatalytic activity of the fibers was studied against the degradation of methylene blue dye under simulated sunlight irradiation and the results of photocatalytic experiments are shown in Figure 10 and Table 2. The first-order apparent reaction rates increase as the amount of g-C3N4 also

. Photocatalytic Activity and Recyclability
The photocatalytic activity of the fibers was studied against the degradation of methylene blue dye under simulated sunlight irradiation and the results of photocatalytic experiments are shown in Figure 10 and Table 2. The first-order apparent reaction rates increase as the amount of g-C3N4 also

Photocatalytic Activity and Recyclability
The photocatalytic activity of the fibers was studied against the degradation of methylene blue dye under simulated sunlight irradiation and the results of photocatalytic experiments are shown in Figure 10 and Table 2. The first-order apparent reaction rates increase as the amount of g-C 3 N 4 also increases according to the following trend k GNT 5% > k GNT 2.5% ≈ k GNT 1% > k TiO2 fibers . Taking into account the similar surface areas, crystal and hydrodynamic sizes and band gap energies of the different fibrous catalysts it can be concluded that interfacial transfer of carriers and the ability for • OH radicals generation are the key-process for the observed photocatalytic performance. This is further supported by the fluorescence measurements that showed the same catalyst sequence with regard to their ability for • OH radicals production.
increases according to the following trend kGNT 5% > kGNT 2.5% ≈ k GNT 1% > kTiO2 fibers. Taking into account the similar surface areas, crystal and hydrodynamic sizes and band gap energies of the different fibrous catalysts it can be concluded that interfacial transfer of carriers and the ability for • OH radicals generation are the key-process for the observed photocatalytic performance. This is further supported by the fluorescence measurements that showed the same catalyst sequence with regard to their ability for • OH radicals production. In order to investigate the recyclability of the prepared photocatalysts, the photocatalytic degradation of MB using the most efficient photocatalyst GNT 5% was studied for three consecutive cycles. According to the experimental results ( Figure 11 and Table 3) the apparent reaction constants determined for the first, second and third catalytic cycle, were 0.0836 min −1 , 0.0797 min −1 and 0.0736 min −1 , respectively. A loss of 12% of its catalytic performance was recorded after the third cycle suggesting a good stability of the catalyst taking also into account some losses during the catalyst recovery process.  In order to investigate the recyclability of the prepared photocatalysts, the photocatalytic degradation of MB using the most efficient photocatalyst GNT 5% was studied for three consecutive cycles. According to the experimental results ( Figure 11 and Table 3) the apparent reaction constants determined for the first, second and third catalytic cycle, were 0.0836 min −1 , 0.0797 min −1 and 0.0736 min −1 , respectively. A loss of 12% of its catalytic performance was recorded after the third cycle suggesting a good stability of the catalyst taking also into account some losses during the catalyst recovery process.
Catalysts 2019, 9, x FOR PEER REVIEW 10 of 15 increases according to the following trend kGNT 5% > kGNT 2.5% ≈ k GNT 1% > kTiO2 fibers. Taking into account the similar surface areas, crystal and hydrodynamic sizes and band gap energies of the different fibrous catalysts it can be concluded that interfacial transfer of carriers and the ability for • OH radicals generation are the key-process for the observed photocatalytic performance. This is further supported by the fluorescence measurements that showed the same catalyst sequence with regard to their ability for • OH radicals production. In order to investigate the recyclability of the prepared photocatalysts, the photocatalytic degradation of MB using the most efficient photocatalyst GNT 5% was studied for three consecutive cycles. According to the experimental results ( Figure 11 and Table 3) the apparent reaction constants determined for the first, second and third catalytic cycle, were 0.0836 min −1 , 0.0797 min −1 and 0.0736 min −1 , respectively. A loss of 12% of its catalytic performance was recorded after the third cycle suggesting a good stability of the catalyst taking also into account some losses during the catalyst recovery process. Figure 11. of GNT 5% fibers for three catalytic cycles. Figure 11. Of GNT 5% fibers for three catalytic cycles. The conduction band (CB) and valence band (VB) potentials for g-C 3 N 4 are −1.3 V and +1.4 V while the respective values for anatase TiO 2 are −0.5 V and +2.7 V, respectively [24,25]. Considering firstly a type II heterojunction the excited electrons from the g-C 3 N 4 CB will be transferred to the less negative TiO 2 CB while the positive holes will be accumulated to the less positive VB of g-C 3 N 4 . However, taking into account the potentials for the oxidation for water (+2.34 V vs. NHE) or OH -(+1.99 V vs. NHE) to • OH radicals it can be concluded that g-C 3 N 4 VB holes cannot oxidize water or OHions to produce • OH radicals. Hydroxy-terephthalic fluorescence measurements in the presence of g-C 3 N 4 supported the above consideration revealing the low production of • OH radicals that can be attributed to the reductive pathway of O 2 following the sequence O 2 •− , HO 2 • , H 2 O 2 . In addition, the composite fibers should present lower photocatalytic activity than the TiO 2 fibers that was not the case [10,17,26]. As a result, type-II photocatalytic mechanism is not probable. On the contrary, the ability of the composite fibers to generation • OH radicals is much greater compared to that of g-C 3 N 4 and higher than that of reference TiO 2 suggesting a Z-scheme mechanism ( Figure 12). In such mechanism, when the electrospun g-C 3 N 4 /TiO 2 fibers are irradiated by the simulated solar light, the photo-generated holes (h + ) were maintained in the VB of TiO 2 and the excited electrons (e -) in the CB of TiO 2 are combined with the holes in the VB of g-C 3 N 4 while electrons are accumulated in the CB of g-C 3 N 4 scavenged by oxygen. As a result, the generated holes in theVB of TiO 2 have the suitable oxidation potential for the generation of • OH radicals while the trapping of e -(CB TiO2 ) by h + (VB g-C3N4 ) induced a better charge separation, thus a higher photocatalytic efficiency.  The conduction band (CB) and valence band (VB) potentials for g-C3N4 are −1.3 V and +1.4 V while the respective values for anatase TiO2 are −0.5 V and +2.7 V, respectively [24,25]. Considering firstly a type II heterojunction the excited electrons from the g-C3N4 CB will be transferred to the less negative TiO2 CB while the positive holes will be accumulated to the less positive VB of g-C3N4. However, taking into account the potentials for the oxidation for water (+2.34 V vs. NHE) or OH -(+1.99 V vs. NHE) to • OH radicals it can be concluded that g-C3N4 VB holes cannot oxidize water or OHions to produce • OH radicals. Hydroxy-terephthalic fluorescence measurements in the presence of g-C3N4 supported the above consideration revealing the low production of • OH radicals that can be attributed to the reductive pathway of O2 following the sequence O2 •− , HO2 • , H2O2. In addition, the composite fibers should present lower photocatalytic activity than the TiO2 fibers that was not the case [10,17,26]. As a result, type-II photocatalytic mechanism is not probable. On the contrary, the ability of the composite fibers to generation • OH radicals is much greater compared to that of g-C3N4 and higher than that of reference TiO2 suggesting a Z-scheme mechanism ( Figure 12). In such mechanism, when the electrospun g-C3N4/TiO2 fibers are irradiated by the simulated solar light, the photo-generated holes (h + ) were maintained in the VB of TiO2 and the excited electrons (e -) in the CB of TiO2 are combined with the holes in the VB of g-C3N4 while electrons are accumulated in the CB of g-C3N4 scavenged by oxygen. As a result, the generated holes in theVB of TiO2 have the suitable oxidation potential for the generation of • OH radicals while the trapping of e -(CBTiO2) by h + (VBg-C3N4) induced a better charge separation, thus a higher photocatalytic efficiency.

Preparation of Electrospun Fibrous TiO 2 /g-C 3 N 4 Heterojunction Photocatalysts
The TiO 2 /g-C 3 N 4 fibers with different g-C 3 N 4 contents were synthesized via the single-nozzle electrospinning technique, in a typical horizontal set up of electrospinning apparatus using a grounded collecting metal plate. Firstly, a PVP (0.125 g) solution in absolute ethanol (2.5 mL) was prepared followed by the addition of TTiP (1 mL) and acetic acid (1 mL) and subsequently stirred by vortex for 5 min to achieve a homogeneous solution. Thereafter, the g-C 3 N 4 powder (1, 2.5 and 5% wt% in respect to the TTiP weight) was added to the solution and the mixture was subjected to bath sonication for 1 h to achieve a fine dispersion of g-C 3 N 4 . Afterward, this mixture was added immediately into a plastic syringe and placed in the syringe-pump (Holliston, MA, USA). Electrospinning was carried out at room conditions using the following parameters: applied voltage 25 kV, tip to collector distance 15 cm and solution feed rate of 1 mLh −1 . Subsequently, the as spun-fibers were calcinated at 550 • C for 3 h. The as-prepared catalysts were named as GNT 1%, GNT 2.5% and GNT 5%, respectively to the % weight of g-C 3 N 4 . TiO 2 fibers were also fabricated by using the similar method in the absence of g-C 3 N 4 and used as a reference material. The g-C 3 N 4 powder was synthesized by thermal condensation process using urea as a precursor compound. Urea was dried in 90 • C for 24 h and then was calcinated in air at 500 • C for 4 h in an aluminum crucible at the heating rate of 10 • C min −1 [4,27] and then, the furnace was cooled naturally to ambient temperature. The obtained yellow color solid was ground well into a powder in an agate mortar.

Characterization
The X-ray diffraction (XRD) patterns of the prepared catalysts were recorded using a Bruker Advance D8 instrument (Billerica, MA, USA) working with Cu-K a (λ = 1.5418 Å) radiation. Diffractograms were scanned from 2θ 10 • to 90 • in steps of 0.02 • and a rate 0.01 • θ sec −1 . The patterns were assigned to crystal phases with the use of the International Center for Diffraction Data (ICCD).
Nitrogen adsorption-desorption isotherms were collected by porosimetry using a Quantachrome Autosorb-1 instrument (Bounton Beach, FL, USA) at 77 K. All the samples (≈0.1 g) were degassed at 423 K for 3 h. Brunauer-Emmet-Teller (BET) method was used at relative pressure between 0.05-0.3, in order to calculate the specific surface area (SSA) of each material. Adsorbed amount of nitrogen at relative pressure P/P o = 0.99 was used in order to calculate the total pore volume (V TOT ). The BJH (Barrett, Joyner and Halenda) method was used to determine the pore size distribution (PSD) of the photocatalysts.
Fiber's morphology and size were observed by scanning electron microscopy (SEM) using a JEOL JSM 5600 instrument (Tokyo, Japan) working at 20 kV. Thermogravimetric analysis (TGA) was performed by STA 449C Netzsch instrument (Selb, Germany) under air atmosphere with a heating rate of 10 • C min −1 from 25 to 800 • C.
A Shimadzu SALD-2300 laser diffraction particle size analyser (Kyoto, Japan) working with dynamic light scattering (DLS) mode was used for catalyst particle size measurements. Suspensions of the nanofibrous photocatalysts were prepared by sonication with a Hielscher UP100H ultrasonic processor (Teltow, Germany) for 10 min.

UV-Vis-DRS Measurements
Diffuse reflectance spectra of the prepared fibrous catalysts were carried out on a Shimadzu 2600 spectrophotometer bearing an IRS-2600 integrating sphere (Kyoto, Japan) in the wavelength range of 200-800 nm at room temperature using BaSO 4 (Nacalai Tesque, extra pure reagent, Kyoto, Japan), as a reference sample.

Determination of • OH Radicals by Fluorescence Measurements
The formation of hydroxyl radicals was studied using terephthalic acid (TA) (98%, Sigma-Aldrich, St. Louis, MO, USA) method. An aqueous solution of NaOH (2 × 10 −3 M, 99% Riedel-de Haën, Seelze, Germany) and TA (5 × 10 −4 M) was prepared and then 20 mg of each photocatalyst powder was added in the photocatalytic reactor and dispersed by sonication for 10 min [28]. Then, the suspensions were irradiated under the same experimental conditions used for the photocatalytic experiments. Aliquots of 4 mL were taken out at different time periods and were centrifuged for 30 min (4400 rpm) to separate the photocatalyst. The intensity of the fluorescence peak at 425 nm (using 310 nm excitation wavelength), which is attributed to 2-hydroxyterephthalic acid (TAOH), was measured by a fluorescence spectrophotometer (Shimadzu RF-5301PC, Kyoto, Japan). It is known, that the amount of generated • OH radicals is proportional to the produced 2-hydroxyterephthalic acid (TAOH). The concentration of • OH was measured by a calibration curve plotting the fluorescence intensity of standard TAOH (TCI, >98%) solutions.

Evaluation of Photocatalytic Performance
The solar simulator apparatus Suntest XLS+ (Atlas, Germany) disposing a xenon lamp (2.2 kW) jacked with special 290 nm cut-off glass filter was used for the photocatalytic experiments. The irradiation intensity was maintained at 500 Wm −2 during the experiments. The photocatalytic activity was evaluated towards the degradation of methylene blue (MB) dye. The photocatalysts (200 mg L −1 ) were sonicated for 10 min in an aqueous solution to obtain a fine suspension. For the photocatalytic experiments 100 mL of MB (5 mg L −1 ) solution were loaded in appropriate Pyrex glass reactor (250 mL) thermostated at ambient conditions (23 ± 1 • C), by water circuit flowing in the double-skin of the reactor and air-flow, under continuous stirring. Before illumination the suspension was magnetically stirred for 30 min to ensure the establishment of adsorption-desorption equilibrium onto the catalyst surface. Samples (5 mL) were withdrawn regularly from the reactor and centrifuged at 4400 rpm (Thermo Scientific, Heraus Megafuge 8, Suzhou, China) immediately for 30 min in order to separate the catalyst particles. The supernatant transparent solution was analysed by UV-Vis spectroscopy (Jasco-V630, Tokyo, Japan) measuring the absorbance at the characteristic wavelength 665 nm using a calibration curve. Relative errors lower than 4.3% were obtained in all cases. For the experiments of catalyst recycling, the materials were recovered by centrifugation, washed twice with distilled water (20 mL) in a vortex, centrifuged again and dried at 383 K for 4 h, finally re-suspended and irradiated using the same experimental conditions.

Conclusions
In summary, g-C 3 N 4 /TiO 2 composite photocatalysts (up to 5% wt of g-C 3 N 4 ) have been successfully prepared with one-step electrospinning process. All the electrospun composite fibers presented mainly anatase crystal phase with small particle sizes, mesoporosity, a response to visible light irradiation and an increased ability to generate • OH radicals. The photocatalytic activity for methylene blue degradation increases as increasing the amount of g-C 3 N 4 coinciding with the trend followed for the • OH radicals generation rate. The composite with 5% wt of g-C 3 N 4 was found as the most efficient photocatalyst with a sufficient stability after three consecutive cycles. A Z-scheme photocatalytic mechanism was proposed for the composite materials.