Behavior of N-doped TiO 2 and N-doped ZnO in Photocatalytic Azo Dye Degradation under UV and Visible Light Irradiation: A Preliminary Investigation

: N-doped TiO 2 (N-TiO 2 ) and N-doped ZnO (N-ZnO) were synthesized utilizing ammonia as a dopant source. The chemico-physical characteristics of synthesized samples were studied by Raman spectroscopy, X-ray diffraction, SEM analysis, N 2 adsorption–desorption at −196 °C, and diffuse reflectance spectroscopy. Compared to undoped samples, the introduction of nitrogen in the semiconductor lattice resulted in a shift of band-gap energy to a lower value: 3.0 eV for N-ZnO and 2.35 eV for N-TiO 2 . The photocatalysts were tested for the degradation of Eriochrome Black T (EBT), which was selected as a model azo dye. Both N-doped semiconductors evidenced an improvement in photocatalytic activity under visible light irradiation (62% and 20% EBT discoloration for N-TiO 2 and N-ZnO, respectively) in comparison with the undoped samples, which were inactive in the presence of visible light. Different behavior was observed under UV irradiation. resulted in a drastic reduction in photocatalytic activity, with only 45% EBT discoloration compared to total color removal obtained with the commercial ZnO sample, suggesting intrinsic limitations for doping of this class of semiconductors.


Introduction
In the last decade, worldwide concerns about environmental and energy-related issues have prompted considerable interest in semiconductor-based heterogeneous photocatalysis [1]. This photocatalytic process can allow sunlight to be used to perform chemical reactions with the aim of degrading highly toxic organic and inorganic substances [2][3][4][5], in addition to the selective synthesis of specific organic compounds under mild conditions [6] and the production of solar fuels (such as hydrogen and methane) [7][8][9]. The most studied semiconductor is TiO2, although promising results have also been obtained in the presence of ZnO [10,11]. However, an important drawback of TiO2 and ZnO semiconductors for photocatalysis is that their band gaps are rather large (3.2 eV [12] and 3.4 eV, respectively) making these photocatalysts inactive under visible light (46% of the solar spectrum). To reduce the energy threshold for photoexcitation, a considerable amount of research has been focused on the doping of semiconductors lattice with both transition metals and non-metal elements. The introduction of dopant elements generates energy between the valence and conduction bands of bare semiconductors, leading to suitable modifications of light absorption properties. Among all the dopant elements, N-doped semiconductors seem to be promising photocatalysts that are active under visible light irradiation [13]. In particular, the doping of TiO2 or ZnO lattice with nitrogen attracted considerable attention because, as reported in the literature [14], the ionic radii of nitrogen and oxygen are similar, and the energy states of N 2p and O 2p are also similar, with low energy formation. In the case of ZnO, nitrogen is proposed to substitute O sites in the crystalline framework [15], whereas for the TiO2 crystalline structure, nitrogen doping can result in nitrogen species, such as substitutional or interstitial nitrogen [16]. The possible formation of NOx species is mostly detected on the surface of semiconductors [17], although the interstitial or substitutional N is preferably present in subsurface layers [17]. Nitrogen has also been reported to simultaneously substitute both O and Ti sites in TiO2 to form Ti1−yO2−xNx+y instead of TiO2−xN [18]. Various methods have been employed for the incorporation of nitrogen in TiO2 or ZnO, either based on chemical reactivity (for example, the sol-gel method [15]) or physical methods (for example, magnetron sputtering [19]). The difference seems to lead, at least in some cases, to photocatalytic materials with somewhat differing properties. Therefore, conflicting results have been reported in the literature in this field. Therefore, one of the key aspects that is still under debate is the chemical nature and the location in the semiconductor lattice of the nitrogen species responsible for photoactivity in visible light. However, papers reporting a direct comparison of the photoactivity of N-doped TiO2 (N-TiO2) and N-doped ZnO (N-ZnO) are still limited in number. Therefore, in this study, we aim to characterize N-TiO2 and N-ZnO photocatalysts from a chemico-physical point of view and preliminarily investigate their photocatalytic behavior under UV and visible light irradiation to assess the influence of N species in various semiconductor crystalline hosts on photocatalytic degradation of Eriochrome Black T (EBT) dye, which was selected as a model azo dye. EBT is widely used in complexometric titrations and to dye silk, wool, and nylon [20]. However, EBT is hazardous (carcinogenic) in the event of skin contact and an irritant in the event of eye contact [21].

Photocatalyst Characterization
The list of the tested photocatalysts is reported in Table 1. UV-Vis diffuse reflectance spectroscopy (UV-Vis DRS) was employed to study the light absorption properties of all the photocatalysts (Figure 1). Band-gap analysis was performed using the Kubelka-Munk function (F(R∞)), and the obtained results are shown in Figure 2 and Table 1.  Table 1), which is an unexpected value for bare TiO2 in the anatase phase. The intensity and width of this absorption depend on the preparation method used for synthesis, as reported by di Valentin et al. [17]. Nevertheless, the spectrum of N-TiO2 is very close to that reported by other authors who synthetized N-doped TiO2 with different methods [22,23]. On the other hand, the optical absorbance behavior of N-ZnO changed slightly with respect to the commercial sample, indicating that doping alters the optical behavior of N-ZnO to a lesser extent than doping of TiO2 with nitrogen. A small decrease in band-gap energy of N-ZnO (3.0 eV) was observed with respect to commercial undoped ZnO (3.15 eV) ( Figure  2b and Table 1), suggesting that the energetic intermediate levels generated between the conduction and valence bands of ZnO do not drastically change the band gap value [24]. Raman spectroscopic analysis was employed to examine the possible variations in Raman active vibrational modes of TiO2 and ZnO due to the presence N into the host lattice ( Figure 3). For both TiO2 and N-TiO2 samples (Figure 3a), all the signals can be assigned to the Raman active modes of the anatase crystal phase [25,26]. Moreover, a slight blue shift of the Raman active mode (Eg(1)) was observed from 144 to 146 cm −1 for the doped sample, clearly indicating the presence of a nonstoichiometric TiO2-x phase [27,28]. For both ZnO and N-ZnO spectra (Figure 3b), a prominent peak was detected at 438 cm −1 , which was associated with the Raman-active dominant mode (E2(H)) of wurtzite ZnO [29]. Furthermore, two less intense Raman bands were observed at 330 and 378 cm −1 and assigned to E2(H)-E2(L) (multi-phonon) and A1 (TO) modes of wurtzite ZnO, respectively [30]. The N-ZnO sample showed three additional broad and weak signals at 275, 504, and 644 cm −1 [31]. It has been argued that the presence of these Raman signals is associated with the presence of nitrogen in the ZnO crystalline framework [31,32]. For the N-ZnO sample, the intensity of the Raman band at 580 cm −1 was stronger than that of the commercial ZnO. According to the available literature on doped ZnO photocatalysts [33][34][35],  Wide-angle X-ray diffraction (WAXD) was used to identify the crystalline phase structure of all the samples (Figure 4). Figure 4a shows the spectrum of N-TiO2 in comparison with TiO2. The analysis demonstrates that doping with nitrogen does not change the crystalline structure of TiO2, as all the detected patterns for both N-TiO2 and TiO2 can be attributed to the anatase phase (JCPS card no. 21-1272) [36]. No other phases, such as   ( 2 2 0) ( 2 1 5)  (1 1 6)   (2 0 4)  (2 1 1)  (1 0 5)   (2 0 0)  (1 1 2) Intensity, a.u. N-TiO2 shows a small shift of the (101) reflection toward a lower diffraction angle (Figure 5a), indicating an enlargement of the cell parameters, which is commonly associated with the formation of an interstitial solid solution (i.e., nitrogen occupies interstitial positions in the TiO2 lattice) [37]. N-ZnO (Figure 5b) evidences a small shift of the (002) peak toward lower angles, which is attributed to the incorporation of N at O sites, indicating that the N ionic radius (0.13 Å) is slightly larger than the O ionic radius (0.12 Å) [38,39]. The WAXD results suggest that in the case of N-TiO2, nitrogen occupies interstitial positions, whereas for N-ZnO, some oxygen sites are replaced by nitrogen.

2, degree
The average crystallite sizes of the photocatalysts were calculated using the Debye-Scherrer equation, considering the (101) reflex for titania in the anatase phase and the (100) reflex for the ZnO wurtzite phase. The obtained values are reported in Table 1.

2, degree
The crystal size of TiO2 and N-TiO2 is almost the same, whereas in the case of N-ZnO, a slight increase was observed.
SEM analysis of N-ZnO and N-TiO2 samples is reported in Figure S1 of Supplementary Material. N-TiO2 is formed by particles of varying size and with a non-regular geometry, whereas N-ZnO is characterized by particles with needle-like morphology. Table 1 reports the specific surface area (SSA) obtained by the BET method for the prepared samples. The SSA of TiO2 slightly decreases from 50 m 2 ·g −1 to 45 m 2 ·g −1 after doping with nitrogen. The SSA value of ZnO and N-ZnO was equal to 5 m 2 ·g −1 and 2 m 2 ·g −1 , respectively. Such results are in agreement with the available literature [40,41]. Figure 6 shows the photocatalytic performances of all the samples toward EBT dye degradation under UV and visible light for TiO2-based photocatalysts (Figure 6a) and ZnO-based photocatalysts (Figure 6b). It is important to note that the photolysis of EBT dye can be negligible under both irradiation conditions [42]. According to the curve decay observed in terms of EBT relative concentration, commercial TiO2 and ZnO are active only under UV irradiation, reaching 56 and 100% EBT discoloration, respectively, after 180 min of irradiation time (Figure 6a,b). The photocatalytic performances of commercial TiO2 and ZnO are extensively discussed and compared in the literature, clearly reporting that ZnO is more active than TiO2 in the degradation of azo dyes using UV or sunlight as the excitation source [43]. It has been argued that the higher photocatalytic activity of ZnO relative to that of TiO2 under UV light may be associated with the increased efficiency of generation and separation of photoinduced electrons and holes [44]. In particular, some authors argued that the intrinsic defects in ZnO crystal lattice are able to effectively trap the charge carriers [45,46]. In particular, Zn interstitials and oxygen vacancies (both positively charged) can enhance the photo-induced redox reactions on ZnO by trapping photo-generated electrons, therefore limiting the recombination phenomena among electrons and holes [47]. Different behavior under UV light was observed when nitrogen was inserted in the crystalline structure of both TiO2 (Figure 6a) and ZnO (Figure 6b).

Photocatalytic Activity Results
N-TiO2 is able to absorb both UV and visible light (Figure 1a), resulting in an enhanced photocatalytic performance compared to commercial TiO2 (Figure 6a), showing the highest photocatalytic activity under both irradiation conditions. With a fixed irradiation time, the EBT relative concentration for N-TiO2 is lower than that observed for TiO2, reaching a discoloration efficiency of 90% and 62% after 180 min of UV and visible light irradiation, respectively. The efficacy of the prepared N-TiO2 was previously observed for the degradation of organic pollutants under visible light irradiation [48][49][50]. This might be due to interstitial nitrogen [16] (as evidenced by the WAXD analysis reported in Figure 5a), which generates intra-gap energy states above the Fermi level, causing a redshift of the N-TiO2 absorption band edge toward the visible region [16,51]. Additionally, Di Valentin et al. underlined the beneficial contribution of oxygen vacancies to the overall visible light photoactivity of N-TiO2 [52]. The oxygen vacancies may act as shallow donors, increasing the carrier density and facilitating the separation and transport of photo-induced carriers [53]. The possible presence of oxygen vacancies was underlined in the Raman results of N-TiO2 sample (Figure 3a). Accordingly, the visible light activity of N-TiO2 could be ascribed to both the presence of impurity states generated in the band gap of TiO2 and the presence of oxygen vacancies induced by nitrogen doping [16,54].
The N-ZnO photocatalyst showed a discoloration efficiency of about 20% under visible light and 45% under UV light (Figure 6b). Although N doping induced photoactivity under visible light, the photocatalytic efficiency of N-ZnO under UV light was significantly lower than that of undoped ZnO, in contrast to the results of the UV light test in the presence of N-TiO2 and TiO2 (Figure 6a).
Parhizgar et al. [55] reported that the presence of defects, such as oxygen vacancies (VO), does not improve the photocatalytic performance of N-ZnO. Instead, the electron transfer from the donor sites (VO) to the acceptor sites (N) determines a shift of the Fermi level toward the conduction band of ZnO, decreasing the mobility of the holes. Although the available literature [56,57] contains some reports of an enhancement in photocatalytic activity, the results in terms of EBT discoloration (Figure 6b) show that doping with nitrogen is not an efficient way to enhance the chemical reactivity of ZnO in the UV region [58]. Moreover, the achieved results could be also attributed to the concentration of N in the ZnO lattice, which could affect the photocatalytic activity. An excess of N was observed to cover most of the surface of ZnO, inhibiting the direct exposure of the ZnO surface to the excitation light, resulting in a reduction in photocatalytic activity [59,60].

Preparation of N-doped Photocatalysts
Commercial anatase TiO2 (PC50, Millennium Chemicals, Hunt Valley, MD, USA) and commercial ZnO (Aldrich, St. Louise, MO, USA) were employed as reference materials and used as received. N-doped TiO2 (N-TiO2) photocatalyst was prepared by sol-gel method. In detail, 25 mL of Ti(OCH(CH3)2)4 (Aldrich, 99%) was added dropwise to 100 mL of NH4OH solution (Carlo Erba, Val-de-Reuil, France, 30 wt%) at 0 °C under continuous stirring. The reaction involves the formation of white precipitate, which was washed with distilled water and centrifuged. The precipitate was then calcined at 450 °C for 30 min to obtain the N-TiO2 sample. N-doped ZnO (N-ZnO) catalyst was prepared by the precipitation method starting with 0.27 g of ZnSO4 (Aldrich, 99%) dissolved in 50 mL of distilled water containing 10 mL of NH4OH solution (Carlo Erba, Val-de-Reuil, France, 28-30%). Then, NaOH solution (4 g of NaOH in 25 mL of distilled water) was slowly added at room temperature to obtain a precipitate. The generated precipitate was then centrifuged, washed, and calcined at 450 °C for 30 min to obtain the N-ZnO sample.

Photocatalyst Characterization
The ultraviolet-visible diffuse reflectance spectra (UV-Vis DRS) of the samples were recorded using a spectrophotometer (Lambda 35, Perkin Elmer, Waltham, MA, USA) equipped with an RSA-PE-20 integrating sphere for reflectance spectroscopy (Labsphere Inc., North Sutton, NH). The optical band-gap values of the photocatalysts were then obtained using the Kubelka-Munk function (F(R∞)) by plotting (F(R∞) * hν) 2 vs. hν (eV) in the case of direct transition or by plotting (F(R∞) * hν) 1/2 vs. hν in the case of indirect transition. SEM microscopy (mod. LEO 420, Assing, Rome, Italy) was used to characterize the morphology of N-TiO2 and N-ZnO samples at an accelerating voltage of 20 kV. The Brunauer-Emmett-Teller (BET) surface area of the photocatalysts was determined by dynamic N2 adsorption measurement at −196 °C with a Costech Sorptometer 1042 after pretreatment for 30 min in an He flow at 150 °C. Raman spectra were obtained with a dispersive microRaman spectrometer (Invia Renishaw, Turin, Italy) with a 541 nm laser in the range 100-800 cm −1 . Wide-angle X-ray diffraction (WAXD) patterns were obtained with an automatic Bruker D8 Advance diffractometer (VANTEC −1 detector) using reflection geometry and nickel-filtered Cu-Kα radiation.

Photocatalytic Activity Tests
Photocatalytic tests were carried out in a batch photoreactor with a cylindrical shape (ID = 2.6 cm, L = 41 cm) equipped with an air distributor device (Qair = 150 cm 3 ·min −1 ) to guarantee the presence of oxygen in the reaction medium. Four UV (main emission wavelength = 365 nm) or visible (emission wavelength in the range of 400-750 nm) lamps (Philips, Amsterdam, The Netherlands, 8 W) were placed around the external body of the photoreactor. The photocatalytic tests were conducted at room temperature using Eriochrome Black T (EBT) aqueous solution with an initial concentration equal to 100 mg·L −1 , a photocatalyst dosage of 3 g·L −1 , and a total volume of solution equal to 100 mL. The suspension was left in dark conditions for 120 min to achieve the adsorption/desorption equilibrium of EBT dye on the photocatalyst surface before switching on the light sources. With varying reaction times, 3 mL of the suspension was withdrawn from the photoreactor and analyzed by a UV-Vis spectrophotometer (Thermo Scientific Evolution 201, Thermo Fisher Scientific, Waltham, MA, USA) to check the reaction advancement. Before the UV-Vis analysis, the suspension was centrifuged in order to remove the photocatalyst particles. In detail, the discoloration of the chosen dye was observed by measuring the maximum absorbance value at 512 nm [61].

Conclusions
In this study, the properties of anatase TiO2 and wurtzite ZnO were modified by the introduction of N-dopant into the respective lattices in order to extend the light absorption to the visible region. The presence of nitrogen in the host crystalline structure evidenced a band-gap shift to a lower value: 3.0 eV for N-ZnO and 2.35 eV for N-TiO2. Raman spectroscopy evidenced the presence of oxygen vacancies for both N-TiO2 and N-ZnO. In addition, WAXD analysis indicated that nitrogen occupies interstitial positions in the TiO2 lattice, whereas in the case of N-ZnO, some oxygen sites are replaced by nitrogen. Both N-doped semiconductors evidenced an improvement in photocatalytic activity under visible light irradiation in comparison with the undoped samples, which were totally inactive. In particular, 62% and 20% EBT discoloration was achieved after 180 min for N-TiO2 and N-ZnO, respectively. However, different behavior was observed under UV irradiation. Whereas N-TiO2 was more photoactive than commercial undoped TiO2, N-ZnO showed a drastic reduction in photocatalytic performance, reaching only 45% EBT discoloration compared to the total color disappearance obtained with commercial ZnO. Therefore, in order to formulate a photocatalyst at high efficiency under direct solar light, such results suggest intrinsic limitations of doping of this class of semiconductors and underline the necessity to develop a visible-light active photocatalyst by N-doping without decreasing the UV-light-driven photocatalytic performances of commercial photocatalysts.

Conflicts of Interest:
The authors declare no conflict of interest.