Ni-Doped Titanium Dioxide Films Obtained by Plasma Electrolytic Oxidation in Refrigerated Electrolytes

Porous crystalline Ni-doped TiO2 films were produced using DC plasma electrolytic oxidation in refrigerated H2SO4 aqueous solutions containing NiSO4. The crystalline phase structure consisted of a mixture of anatase and rutile, ranging from ~30 to ~80 wt % rutile. The oxide films obtained at low NiSO4 concentration showed the highest photocurrent values under monochromatic irradiation in the UV-vis range, outperforming pure TiO2. By increasing NiSO4 concentration above a threshold value, the photoelectrochemical activity of the films decreased below that of undoped TiO2. Similar results were obtained using cyclic voltammetry upon polychromatic UV-vis irradiation. Glow discharge optical emission spectrometry (GD-OES) analysis evidenced a sulfur signal peaking at the TiO2/Ti interface. XPS spectra revealed that oxidized Ni2+, S4+ and S6+ ions were included in the oxide films. In agreement with photocurrent measurements, photoluminescence (PL) spectra confirmed that less intense PL emission, i.e., a lower electron-hole recombination rate, was observed for Ni-doped samples, though overdoping was detrimental.


Introduction
For several decades the two main drawbacks of titanium dioxide, i.e., its relatively high band gap (3.0-3.2 eV) and low quantum efficiency in photocatalytic reactions, hindered the development of practical applications of photocatalysis. In order to tune the band gap and regulate the electronic activity of TiO 2 , several strategies have been explored such as sensitization with dyes [1,2] and graphene [3,4], the synthesis of nanocomposites [5,6] and doping using metals [7] or non-metals [8,9].
Nickel is among the metal ions investigated for potential doping of TiO 2 . Based on the calculated total density of states of pure TiO 2 and Ni-TiO 2 with/without oxygen vacancies, Ni is expected to introduce new defect states within the band-gap and to strengthen the intra-band states generated by intrinsic oxygen vacancies due to hybridization of the Ni 3d orbitals with the intrinsic defect states [10,11]. Ni-doping also showed enhanced adsorption in the visible region [12] and photocatalytic

Preparation of Ni-Doped TiO 2 Films
Ni-doped TiO 2 films were prepared using PEO of commercially pure (Grade I) titanium sheets (Industrie DeNora S.p.A., Milano, Italy). Prior to PEO, Ti specimens were etched in a dilute HF aqueous solution. PEO was conducted in DC mode at 150 V for 10 min processing time. During PEO, the maximum current was set at 10 A. The anode was a titanium sheet with 6 cm 2 exposed area, while the cathode was a titanium mesh. The electrolyte consisted of aqueous solutions containing 1.5 M H 2 SO 4 and 0-0.6 mM NiSO 4 , having a total volume of 1 L. During PEO, the electrolyte temperature was set at −5 • C by means of a cryostat (HAAKE D10, Thermo Electron Corp., Karlsruhe, Germany). After PEO, the samples were rinsed with water and dried in an air stream. Reference samples consisting of pure anatase and pure rutile were obtained using PEO at a constant voltage of 120 V and 180 V, respectively. All samples were prepared in duplicate. The obtained films were labeled from S0 to S6 as a function of the NiSO 4 concentration in the electrolyte, as specified in Table 1. Similarly, pure anatase and pure rutile samples were labeled as SA and SR, respectively.

Characterization of TiO 2 Photoanodes
The surface morphology of the TiO 2 films was investigated using scanning electron microscopy (SEM, EVO 50, Carl Zeiss Jena GmbH, Jena, Germany). Film thickness and in-depth elemental composition were determined using glow discharge optical emission spectrometry (GD-OES) using a Spectrum GDA750 analyzer (SPECTRO Analytical Instruments Inc., Kleve, Germany) operated at 700 V in argon atmosphere at 230 Pa. Energy-dispersive X-ray spectroscopy (EDX) and X-ray photoemission spectroscopy (XPS) were also carried out to assess the elemental composition. Both techniques are characterized by similar detection limits (in the 0.1-1 at % range [41,42]), nonetheless XPS is a stand-alone tool that provides higher sensitivity and accuracy for the elemental compounds analysis. XPS was performed in a dedicated ultra-high vacuum system (base pressure in the low 10 −8 Pa) by exciting electrons with an Al-Kα source (hν = 1486.6 eV) and analyzing them with a 150 mm hemispherical analyzer from SPECS GmbH (Berlin, Germany) [43]. We accounted for possible charging effects by setting the peak from adventitious carbon to 285 eV [44].
X-ray diffraction (XRD) patterns were acquired at the scanning rate of 2.5 • min −1 with CuK α1 radiation in the 20 • -60 • 2θ range by means of a PW1830 diffractometer (Malvern Panalytical Ltd., Malvern, UK and Almelo, The Netherlands) operating at 40 kV voltage and 40 mA filament current. The XRD patterns were indexed according to the powder diffraction files released by the International Center for Diffraction Data (Newtown, PA, USA) for titanium (PDF 44-1294), anatase (PDF 21-1272) and rutile phases (PDF 21-1276). The weight fraction of anatase (f A ) was calculated according to Equation (1) [45], where I R is the intensity of the (110) rutile reflection and I A is the intensity of the (101) anatase reflection.
UV-vis-NIR diffuse reflectance spectra (DRS) were recorded in the 220-2600 nm range with a UV3600 Plus spectrophotometer from Shimadzu Corp. (Kyoto, Japan) equipped with an ISR-603 integrating sphere. The band gap was calculated based on the UV-vis reflectance spectra after Kubelka-Munk conversion using the Tauc plot method [46]. Photoluminescence (PL) spectra were recorded with an FLS980 spectrofluorometer (Edinburgh Instrument Ltd., Livingston, UK) in the range starting from 20 nm above the excitation wavelength (300 and 380 nm, respectively) up to 850 nm.
Electrochemical impedance spectroscopy (EIS) measurements were performed in the dark and in 1.0 M NaOH aqueous solutions at the open circuit potential (OCP) of each specimen. The amplitude perturbation was 10 mV and the frequency ranged from 300 kHz to 10 mHz using a ModuLab ® XM ECS high-performance potentiostat/galvanostat system (Solartron Analytical, XM PSTAT 1 MS/s, Ametek Inc., Berwyn, IL, USA). The instrument was coupled with a frequency response analyzer (Solartron Analytical, XM FRA 1MHz, Ametek Inc., Berwyn, IL, USA) for AC measurements. The EIS results were fitted with a Randles-type equivalent circuit with Zview software (Ametek Inc., Berwyn, IL, USA).
The photocurrent density as a function of the irradiation wavelength was measured in a 1.0 M NaOH aqueous solution using an optical bench equipped with a 300 W Xe lamp (Quantum Design Europe GmbH, Darmstadt, Germany), a monochromator (Omni-λ 150, Quantum Design Europe GmbH, Darmstadt, Germany), a shutter (SC10, Thorlabs Inc., Newton, MA, USA) and a homemade three electrode two compartment Plexiglas cell together with an optical Pyrex glass window. A PEO TiO 2 photoelectrode was used as working electrode, and a platinum foil and a saturated calomel electrode (SCE) were used as counter and reference electrodes, respectively. Electrochemical measurements were conducted at room temperature on both biased and unbiased TiO 2 photoelectrodes using a 2549 potentiostat/galvanostat from Amel S.r.l. (Milano, Italy) and a DMM4040 digital multimeter from Tektronix (Beaverton, OR, USA). The incident wavelength values ranged from 250 to 450 nm with a 2 nm step and a 4 s per step dwell time. The incident light power was measured using a calibrated Thorlabs S130VC photodiode connected to a PM200 power meter (Thorlabs Inc., Newton, MA, USA) placed at exactly the same distance as the TiO 2 electrode, with the Pyrex window in between to account for the transmittance of the cell window. The incident photon-to-current efficiency (IPCE) at each wavelength was calculated using the following Equation: where h (kg m 2 s −1 ) is the Planck constant, c (m s −1 ) is the speed of light, e (C) is the electron charge, I (A m −2 ) is the steady-state photocurrent density, P (W m −2 ) is the light intensity and λ (m) is the incident wavelength. Linear sweep voltammetry (LSV) was carried out in a three electrodes cell containing 4 mM KCl aqueous solution. LSV was conducted from −0.5 to 1 V vs. SCE at a scan rate of 10 mV s −1 and room temperature. The working electrode was a titanium dioxide sheet obtained using PEO. The counter electrode was platinum foil, and the reference electrode was a saturated calomel electrode (SCE). The photocurrent was calculated as difference between the current response in the dark and under irradiation, as a function of the polarization potential. For the measurements, a 2549 potentiostat/galvanostat (Amel S.r.l., Milano, Italy) was used. The irradiation was achieved by means of a polychromatic commercial Xenon lamp. The irradiance of the samples was 100 mW cm −2 , as measured by a calibrated Thorlabs S130VC photodiode connected to a Thorlabs PM200 power meter placed at exactly the same distance as the investigated films. The exposed area during LSV was 5 cm 2 . All measurements were repeated twice. Figure 1 shows the SEM micrographs and EDX spectra of sample S0 ( Figure 1a) and S6 (Figure 1b). SEM analysis revealed a surface morphology typical of PEO coatings. Indeed, they are homogeneous and porous, the pores having sub-micrometric dimensions and a branched structure. Apparently, the concentration of Ni sulfate in the electrolyte did not affect the morphology of the obtained film. Based on the EDX elemental analysis, the TiO 2 films consisted of Ti and O elements. No emission from Ni or S was detected; however, a very low concentration of these species could not be ruled out.

Morphology and Composition
The GD-OES in-depth composition profiles of samples S0 and S6 are shown in Figure 2a,b, respectively. In addition to the expected signals for titanium and oxygen elements, both samples revealed a signal corresponding to sulfur, peaking in proximity of the TiO 2 /Ti interface. This signal showed a wide tail spreading across the film thickness and barely reached the TiO 2 surface. The presence of sulfur can be reasonably ascribed to sulfuric acid contained in the electrolyte, while the shape of the sulfur signal might be reasonably explained by blending of the oxide layer taking place during PEO.
Indeed, the intense sparking occurring at the operating conditions locally melted the oxide, probably moving sulfur-containing species initially included in the TiO 2 film up to the surface of the oxide. Unexpectedly, in sample S6 the Ni element was not detected. Film thickness, as assessed using the GD-OES composition profiles, was~3.3 µm regardless of the Ni concentration in the electrolyte. Figure 1 shows the SEM micrographs and EDX spectra of sample S0 ( Figure 1a) and S6 ( Figure  1b). SEM analysis revealed a surface morphology typical of PEO coatings. Indeed, they are homogeneous and porous, the pores having sub-micrometric dimensions and a branched structure. Apparently, the concentration of Ni sulfate in the electrolyte did not affect the morphology of the obtained film. Based on the EDX elemental analysis, the TiO2 films consisted of Ti and O elements. No emission from Ni or S was detected; however, a very low concentration of these species could not be ruled out. The GD-OES in-depth composition profiles of samples S0 and S6 are shown in Figure 2a and b, respectively. In addition to the expected signals for titanium and oxygen elements, both samples revealed a signal corresponding to sulfur, peaking in proximity of the TiO2/Ti interface. This signal showed a wide tail spreading across the film thickness and barely reached the TiO2 surface. The presence of sulfur can be reasonably ascribed to sulfuric acid contained in the electrolyte, while the shape of the sulfur signal might be reasonably explained by blending of the oxide layer taking place during PEO. Indeed, the intense sparking occurring at the operating conditions locally melted the oxide, probably moving sulfur-containing species initially included in the TiO2 film up to the surface A more detailed elemental investigation was performed using XPS. The survey scans of samples S0 and S6 are shown in Figure 3a, where the binding energy (BE) position of photoemission signals from Ni, O, Ti, C and S is also highlighted. The observed C 1s signal is attributed to adventitious carbon. The BE position of O 1s and Ti 2p3/2 peaks is consistent with the values reported in the literature for TiO2 [47,48]. The presence of Ni in sample S6 is testified by the detection of a small photoemission signal from the Ni 2p3/2 orbital, as shown in the detailed scan of Figure 3b. The BE position of the Ni feature (about 856 eV) is typical of oxidized Ni 2+ species [16]. As shown in the inset of Figure 3d, the intensity of photoemission signal from Ni 2p3/2 increases in the order S0 < S2 < S4 < S6, thus demonstrating that the concentration of NiSO4 impacts on the amount of Ni ions included in the oxide films. The S 2p region ( Figure 3c) shows a feature located at a BE of about 169 eV, which is compatible with the presence of photoemission signals from both tetravalent S 4+ and hexavalent S 6+ sulfur (at the very close BE of 168.8 eV and 169.5 eV, respectively), as also reported in literature [49]. Considering these two possible contributions, the strongest peak at 169.5 eV is usually assigned to SO4 2− groups adsorbed on the TiO2 surface [50]. This assignment is reasonable in the present case, considering that PEO of titanium was carried out in sulfuric acid solutions. As for the tetravalent sulfur signal, S 4+ species can be incorporated into TiO2 either interstitially or at the Ti 4+ lattice site [51]. A more detailed elemental investigation was performed using XPS. The survey scans of samples S0 and S6 are shown in Figure 3a, where the binding energy (BE) position of photoemission signals from Ni, O, Ti, C and S is also highlighted. The observed C 1s signal is attributed to adventitious carbon. The BE position of O 1s and Ti 2p 3/2 peaks is consistent with the values reported in the literature for TiO 2 [47,48]. The presence of Ni in sample S6 is testified by the detection of a small photoemission signal from the Ni 2p 3/2 orbital, as shown in the detailed scan of Figure 3b. The BE position of the Ni feature (about 856 eV) is typical of oxidized Ni 2+ species [16]. As shown in the inset of Figure 3d, the intensity of photoemission signal from Ni 2p 3/2 increases in the order S0 < S2 < S4 < S6, thus demonstrating that the concentration of NiSO 4 impacts on the amount of Ni ions included in the oxide films. The S 2p region ( Figure 3c) shows a feature located at a BE of about 169 eV, which is compatible with the presence of photoemission signals from both tetravalent S 4+ and hexavalent S 6+ sulfur (at the very close BE of 168.8 eV and 169.5 eV, respectively), as also reported in literature [49]. Considering these two possible contributions, the strongest peak at 169.5 eV is usually assigned to SO 4 2− groups adsorbed on the TiO 2 surface [50]. This assignment is reasonable in the present case, considering that PEO of titanium was carried out in sulfuric acid solutions. As for the tetravalent sulfur signal, S 4+ species can be incorporated into TiO 2 either interstitially or at the Ti 4+ lattice site [51]. Electrochemical impedance spectroscopy tests were carried out to investigate the electrochemical properties of the oxide films. The resulting Nyquist plots shown in Figure 4 exhibit a typical semicircle shape. The semicircle obtained for sample S2 is smaller than that recorded for samples S0, S4 and S6, suggesting lower charge transfer resistance allowing a more efficient transport and separation of electron-hole pairs. Electrochemical impedance spectroscopy tests were carried out to investigate the electrochemical properties of the oxide films. The resulting Nyquist plots shown in Figure 4 exhibit a typical semicircle shape. The semicircle obtained for sample S2 is smaller than that recorded for samples S0, S4 and S6, suggesting lower charge transfer resistance allowing a more efficient transport and separation of electron-hole pairs. Surfaces 2020, 3 FOR PEER REVIEW 7

Crystal Structure and Band-Gap Determination
XRD results on doped and undoped TiO2 films are shown in Figure 5a. The TiO2 films were mainly crystalline in structure and consisted of a mixture of anatase and rutile allotropic phases, even though the presence of a minor amorphous component in the TiO2 film structure cannot be completely ruled out. The XRD peaks at 2θ ≈ 25.31°, 48.06° and 55.11° were attributed to the anatase phase, those at 2θ ≈ 27.44°, 36.09°, 41.26° and 54.33° to the rutile phase, and finally those at 2θ ≈ 35.06°, 38.40°, 40.17° and 53.01° to the Ti substrate. No Ni-containing phases were detected in the PEO layers. As shown in Figure A1, no lattice distortion was observed by increasing the NiSO4 concentration. Indeed, substitutional Ni ions at Ti 4+ sites are not expected to induce lattice distortion as they have similar ionic radii, i.e., 0.61 and 0.69 Å for Ti 4+ and Ni 2+ , respectively. In agreement with literature, at the highest nickel sulfate concentration (S6) the rutile reflections centered at 2θ ≈ 27.44° and 36.09 became stronger, indicating that a higher Ni-doping favors the formation of the rutile phase [15]. The weight percentage of anatase and rutile phases obtained using Equation (1) is reported in Table 1. The rutile percentage increased from ~30% to ~80% at increasing nickel sulfate concentration. Figure  5b shows the UV-vis-NIR diffuse reflectance spectra of the investigated samples as Tauc-plots of the Kubelka-Munk transform. As expected, the calculated band-gap values reported in Table 1 are in good agreement with the corresponding phase composition.

Crystal Structure and Band-Gap Determination
XRD results on doped and undoped TiO 2 films are shown in Figure 5a. The TiO 2 films were mainly crystalline in structure and consisted of a mixture of anatase and rutile allotropic phases, even though the presence of a minor amorphous component in the TiO 2 Figure A1 (Appendix A), no lattice distortion was observed by increasing the NiSO 4 concentration. Indeed, substitutional Ni ions at Ti 4+ sites are not expected to induce lattice distortion as they have similar ionic radii, i.e., 0.61 and 0.69 Å for Ti 4+ and Ni 2+ , respectively. In agreement with literature, at the highest nickel sulfate concentration (S6) the rutile reflections centered at 2θ ≈ 27.44 • and 36.09 became stronger, indicating that a higher Ni-doping favors the formation of the rutile phase [15]. The weight percentage of anatase and rutile phases obtained using Equation (1) is reported in Table 1. The rutile percentage increased from~30% to~80% at increasing nickel sulfate concentration. Figure 5b shows the UV-vis-NIR diffuse reflectance spectra of the investigated samples as Tauc-plots of the Kubelka-Munk transform. As expected, the calculated band-gap values reported in Table 1 are in good agreement with the corresponding phase composition. Figure A2 shows the photocurrent density values measured for all samples as a function of the irradiation wavelength and under a 0.6 V vs. SCE applied bias. The corresponding light power is also reported on the right y-axis of the same figure. The corresponding IPCE curves calculated using Equation (2) are plotted in Figure 6a. The investigated samples exhibit a bimodal curve. According to literature [52,53], the absorption peaks around 325 and 380 nm can be attributed to the anatase and rutile phases, respectively. Thus, in the present case the curve shapes and the relative intensity of the two maxima well reflect the crystalline structure of the films. The highest IPCE values were observed for sample S2, reaching 55% depending on the irradiation wavelength. Samples S0 and S4 showed very similar IPCE values, while the lowest values were measured for S6, which mainly consisted of rutile phase. The photocurrent onset is below 414 nm, in partial agreement with the calculated band gap reported in Table 1.

Photocurrent and IPCE
became stronger, indicating that a higher Ni-doping favors the formation of the rutile phase [15]. The weight percentage of anatase and rutile phases obtained using Equation (1) is reported in Table 1. The rutile percentage increased from ~30% to ~80% at increasing nickel sulfate concentration. Figure  5b shows the UV-vis-NIR diffuse reflectance spectra of the investigated samples as Tauc-plots of the Kubelka-Munk transform. As expected, the calculated band-gap values reported in Table 1 are in good agreement with the corresponding phase composition.   As the potential onset is representative of the conduction band edge energy, the shift towards more positive values (i.e., lower energies) would partially account for the correspondingly smaller band-gap energy values. potential onset shifted towards more positive values for Ni-doped TiO2 films compared to undoped films, except for sample S2. The CBedge values were −0.28 VSCE, pH=7 in S0, −0.31 VSCE, pH=7 in S2, −0.23 VSCE, pH=7 in S4 and in S6, which correspond to −0.45 VNHE, pH=0, -0.48 VNHE, pH=0 and −0.40 VNHE, pH=0, respectively. As the potential onset is representative of the conduction band edge energy, the shift towards more positive values (i.e., lower energies) would partially account for the correspondingly smaller band-gap energy values.
Overall, S2 shows a better photocurrent response, both as a function of the irradiation wavelength and under continuous polychromatic irradiation. In addition to the optimum crystalline phase composition, this might also be partially attributed to the lower charge transfer resistance estimated on the basis of the Nyquist curves shown in Figure 4.

Photoluminescence Spectroscopy
In order to achieve a better insight on the effectiveness of trapping, migration and transfer of charge carriers in the investigated films, the photoluminescence (PL) spectra of some representative samples were acquired upon excitation at 300 nm. They are reported in Figure 7a, together with the spectra of PEO TiO2 films consisting in either pure anatase or almost pure rutile. The first emission peak at 354 nm (3.51 eV), observed for all samples, can be attributed to direct recombination with holes of hot electrons excited in the conduction band (CB), i.e., electrons not fully relaxed to the bottom of the CB. The second peak at 388 nm (3.19 eV) reflects the recombination between electrons at the bottom of the CB and holes in the valence band (VB) of the anatase phase and was detected for all samples, in agreement with the crystalline phase composition. The other emission signals must be necessarily ascribed to electronic transitions involving intra-band energy states. The peak at ~555 nm (2.23 eV) measured for all samples with the exception of sample SR, might be tentatively attributed to a transition from the CB to mid-gap states corresponding to oxygen vacancies generated by the replacement of S 6+ for Ti 4+ [54]. The broad emission band observed between 400 and ~700 nm (i.e., 3.10 eV ÷ 1.77 eV) can be attributed to oxygen vacancies and/or defects. In fact, oxygen vacancies and Overall, S2 shows a better photocurrent response, both as a function of the irradiation wavelength and under continuous polychromatic irradiation. In addition to the optimum crystalline phase composition, this might also be partially attributed to the lower charge transfer resistance estimated on the basis of the Nyquist curves shown in Figure 4.

Photoluminescence Spectroscopy
In order to achieve a better insight on the effectiveness of trapping, migration and transfer of charge carriers in the investigated films, the photoluminescence (PL) spectra of some representative samples were acquired upon excitation at 300 nm. They are reported in Figure 7a, together with the spectra of PEO TiO 2 films consisting in either pure anatase or almost pure rutile. The first emission peak at 354 nm (3.51 eV), observed for all samples, can be attributed to direct recombination with holes of hot electrons excited in the conduction band (CB), i.e., electrons not fully relaxed to the bottom of the CB. The second peak at 388 nm (3.19 eV) reflects the recombination between electrons at the bottom of the CB and holes in the valence band (VB) of the anatase phase and was detected for all samples, in agreement with the crystalline phase composition. The other emission signals must be necessarily ascribed to electronic transitions involving intra-band energy states. The peak at~555 nm (2.23 eV) measured for all samples with the exception of sample SR, might be tentatively attributed to a transition from the CB to mid-gap states corresponding to oxygen vacancies generated by the replacement of S 6+ for Ti 4+ [54]. The broad emission band observed between 400 and~700 nm (i.e., 3.10 eV ÷ 1.77 eV) can be attributed to oxygen vacancies and/or defects. In fact, oxygen vacancies and defects can bind the electrons photoexcited in the CB to form excitons, thus increasing the PL emission [55].
However, based on first-principle band calculations, the presence in the same region of additional intra-band states due to Ti 1-x S x O 2 deriving from substitutional tetravalent S 4+ species cannot be ruled out [56]. S-doping involving both S 6+ and S 4+ species would also be in agreement with the XPS scans shown in Figure 3. According to DFT calculations reported in literature, Ni-doping also induces intra-band Ni 3d states generated by hybridization with the O 2p orbitals [57]. Ni ions can be included in the oxide lattice sites as a substitutional defect, either as Ni 2+ or as Ni 3+ , where Ni 2+ can introduce donor levels above the VB, while Ni 3+ can trap photopromoted electrons [12]. The PL intensity of sample S2 is significantly lower than that of the undoped TiO 2 film, while the contrary was obtained for sample S6. This confirms that doping can effectively suppress the recombination of photogenerated electrons and holes tentatively due to a better oxide stoichiometry, though above a certain threshold value it can have a detrimental effect probably due to charge unbalance [58].
PL emission spectra acquired upon irradiation at 380 nm are also reported (Figure 7b). In this case it is even clearer that by Ni-doping the PL emission decreased with respect to pure TiO2 and that overdoping (sample S6) was somehow detrimental. This is in agreement with literature [59], where it is reported that an excess of dopant corresponds to an excess of structure defects which can play a detrimental role in favoring the undesired electron-hole recombination process.
It is well-known that the PL emission results from the recombination of excited electrons and holes and that high PL intensity corresponds to a high recombination rate. Interestingly, the PL results are in good agreement with the IPCE data reported in Figure 6a. Indeed, higher photocurrent is expected for films showing lower PL emission, i.e., lower electron-hole recombination rate. In the case of the investigated TiO2 films, the optimum doping can be obtained with 0.2 mM NiSO4 in the electrolytic bath.

Conclusions
Crystalline Ni-doped TiO2 films were successfully obtained using PEO carried out in DC mode in refrigerated electrolytes containing 1.5 M H2SO4 and NiSO4 in the 0-0.6 mM range, as a source of nickel. The oxide showed a branched and sub-micrometric porous surface morphology, typical of PEO oxides. The as-prepared oxides were crystalline and mainly consisted of a mixture of anatase and rutile. By adding NiSO4 to the electrolytic bath, the relative amount of anatase was initially only marginally affected and not correlated to the electrolyte composition, although at the highest NiSO4 concentration value a clear stabilization of the rutile phase occurred. In addition to the expected oxygen and titanium signals, GD-OES in-depth profile analysis revealed a sulfur signal peaking in proximity of the TiO2/Ti interface. XPS analysis demonstrated that S-containing species consisted of both substitutional S 4+ ions and S 6+ ions, probably due to SO4 2− adsorbed groups. XPS analysis also evidenced the presence of oxidized Ni 2+ ions, the intensity of the corresponding photoemission signal PL emission spectra acquired upon irradiation at 380 nm are also reported (Figure 7b). In this case it is even clearer that by Ni-doping the PL emission decreased with respect to pure TiO 2 and that overdoping (sample S6) was somehow detrimental. This is in agreement with literature [59], where it is reported that an excess of dopant corresponds to an excess of structure defects which can play a detrimental role in favoring the undesired electron-hole recombination process.
It is well-known that the PL emission results from the recombination of excited electrons and holes and that high PL intensity corresponds to a high recombination rate. Interestingly, the PL results are in good agreement with the IPCE data reported in Figure 6a. Indeed, higher photocurrent is expected for films showing lower PL emission, i.e., lower electron-hole recombination rate. In the case of the investigated TiO 2 films, the optimum doping can be obtained with 0.2 mM NiSO 4 in the electrolytic bath.

Conclusions
Crystalline Ni-doped TiO 2 films were successfully obtained using PEO carried out in DC mode in refrigerated electrolytes containing 1.5 M H 2 SO 4 and NiSO 4 in the 0-0.6 mM range, as a source of nickel. The oxide showed a branched and sub-micrometric porous surface morphology, typical of PEO oxides. The as-prepared oxides were crystalline and mainly consisted of a mixture of anatase and rutile. By adding NiSO 4 to the electrolytic bath, the relative amount of anatase was initially only marginally affected and not correlated to the electrolyte composition, although at the highest NiSO 4 concentration value a clear stabilization of the rutile phase occurred. In addition to the expected oxygen and titanium signals, GD-OES in-depth profile analysis revealed a sulfur signal peaking in proximity of the TiO 2 /Ti interface. XPS analysis demonstrated that S-containing species consisted of both substitutional S 4+ ions and S 6+ ions, probably due to SO 4 2− adsorbed groups. XPS analysis also