Introduction

Semiconductor photocatalysis and photoelectrocatalysis have been the subject of many research studies due to the wide range of applications that have been explored for the elimination of hazardous pollutants in air, soil, and water (Han et al. 2009; Zhao et al. 2007; Baea et al. 2008). TiO2 is particularly popular due to its advantageous characteristics such as high photo activity, thermal and chemical stability, low cost and non-toxicity (Castro et al. 2009). On the other hand, the physical and chemical properties of TiO2 can be controlled by its particle size, morphology, and crystalline phase (anatase, rutile, and brookite) (Kim et al. 2008).

Nevertheless, the practical applications of TiO2 are limited by two main issues. While one of them is the low quantum yield that results from the rapid recombination of photo-generated electrons and holes, the other is related to the semiconductor band gap (3.2 eV) that only absorbs effectively 3–4% of the solar energy spectrum (Tian et al. 2009).

For these reasons, several studies have been carried out in order to improve the photocatalytic properties of TiO2. The main approach consists on doping the TiO2 material with metals such as Fe, Co, Ni, Mn, Au, Ag, and Pt, or with oxides such as WO3 (Li et al. 2008; Gonçalves et al. 2006; Khaleel and Al-Nayli 2008; Tryba 2008; Glaspell and Manivannan 2005; Sobana et al. 2008; Muruganadham et al. 2006; Tiana et al. 2008). Metals either doped or deposited on TiO2 are expected to show several effects on the photocatalytic activity due to different mechanisms. For instance, the metal may (i) enhance the electron–hole separation by acting as electron scavenger and (ii) extend the light absorption of the semiconductor material into the visible range. In this way, it has been suggested that a transition metal with a proper oxidation state replaces some of the Ti4+ centers in the lattice producing an impurity state that reduces the band gap (Kudo et al. 2007; Khan and Kim 2009). In spite of a large quantity of literature dealing with the effect of dopants on the chemical and physical properties of TiO2, there is very little information available on the location of the dopant within the semiconductor structure and its local environment.

In order to obtain TiO2 doped with transition metals, several methods have been developed, being the sol–gel method (Garzella et al. 2003; Tian et al. 2009; Piera et al. 2003; Pacheco et al. 2004; Gonçalves et al. 2006; Facchin et al. 2000; Eshaghi et al. 2011; Kisand et al. 2010) the most frequently employed. However, the synthesis of doped TiO2 starting from metallic titanium mixed with other metals has not been thoroughly explored.

On the other hand, to the best of our knowledge, the preparation of electrophoretically deposited films of TiO2 modified with cobalt or nickel has not been reported. This approach is advantageous because electrophoretic deposition allows the control of the film thickness, porosity, and roughness factor (Esquivel et al. 2009, Manriquez and Godínez 2007) in a short period of preparation time as compared to spin coating, immersion, or impregnation deposit methods (Kim et al. 2007; Bengtsson et al. 2009; Suriye et al. 2007). Additionally, semiconductor films supported on optically transparent electrodes (OTE) avoid the filtration process that is necessary when photocatalytic degradation tests are carried out in suspension (Barakat et al. 2004; Li et al. 2008; Chu et al. 2008, 2009; Esquivel et al. 2009; Selculi et al. 2004).

In this context, the aim of this study is to prepare and characterize TiO2 films on OTE doped with Co or Ni using a non-common method that starts from metallic titanium. The obtained deposits were tested for a preliminary photoelectrocatalytic degradation of a model dye (Orange II) in aqueous medium. Additionally, Raman spectroscopy and X-ray diffraction analysis led us to propose the cation location within the TiO2 lattice.

Materials and methods

M:TiO2 synthesis

The synthesis method is a modification of those reported previously by Martín et al. (2003). As titanium source, metallic titanium powder (99.7%, Aldrich) was used. A precursor solution was obtained by dissolving the titanium powder in hydrochloric acid (HCl, 36.5–38%, J.T. Baker) under stirring conditions for 3 h at the atmospheric boiling point (1 g Ti/20 mL HCl). Once the titanium tetrachloride was obtained in this reaction mixture, nitric acid (10 vol.%) solution was added drop by drop until a color change was visible (from purple to light yellow). This step indicates the complete oxidation from Ti3+ to Ti4+. Then, the solution was cooled and the dopant salt was added, nickel (as Nickel acetate 99%, Ni(CH3COO)2·XH2O, Alfa Aesar) or cobalt (as Cobalt acetate 98%, Co(CH3COO)3·4H2O, Caledon) at different weight percentages (wt%). For both cations, the wt% used were 5, 20, 40, and 50. Ammonium hydroxide (NH3 content 28–30%, Sigma-Aldrich) was used as precipitating agent. The powder was filtered and washed several times with deionized water. The TiO2 modified powder was freeze dried (Free zone 2.5 LABCONCO). The synthetic process with its corresponding chemical equations is presented below (Eqs. 1–3).

$$ {\text{Ti}}^{0} + 6{\text{HCl}} \to {\text{TiCl}}_{4} + 2{\text{HCl}} + 4{\text{H}}^{ + } $$
(1)
$$ {\text{TiCl}}_{4} + {\text{M}}^{2 ,3 + } \to {\text{M}}^{2 ,3 + } - {\text{TiCl}}_{4} + {\text{HCl}} $$
(2)
$$ {\text{M}}^{2 ,3 + } - {\text{TiCl}}_{4} + 2{\text{NH}}_{4} {\text{OH}} \to {\text{M:TiO(OH)}}_{2} + 2{\text{HCl}} + 2{\text{NH}}_{4} {\text{Cl}}$$
(3)

Electrophoretic deposition of M:TiO2 powder on ITO electrodes

M:TiO2 films were deposited on ITO glass (Indium Tin Oxide, conductive substrate, 1 cm2 geometric area) using the electrophoretic method (Manriquez and Godínez 2007). In this way, the glass substrate was immersed for 60 s at room temperature in a previously sonicated colloidal suspension (1 g M:TiO2 powder in 10 mL Ethyl Alcohol absolute, J.T. Baker), applying a potential difference of 4 V between a stainless steel sheet and the ITO electrode (Esquivel et al. 2009). The modified substrates were then placed in an oven at 550 °C for 3 h in order to obtain the anatase phase of the modified TiO2 film.

Modified TiO2 characterization and Rietveld Refinement

Surface morphology analysis of the deposited materials was carried out using scanning electron microscopy (SEM) with a JEOL JSM-5400L microscope. The accelerating voltage employed was 15 kV.

Energy dispersive X-ray spectroscopy (EDX) allowed a rough surface elemental analysis and an approximate atomic relationship between Ti4+ and the M2,3+ species (EDX Oxford Inca X-Sight coupled to a JEOL JSM-6060 LV). Band gap values were determined from UV–Vis diffuse reflectance measurements (Ocean Optics Spectrophotometer) applying the Kubelka–Munk theory (Janczarek et al. 2007). X-ray diffraction analysis (XRD) was performed using a Bruker-AXS D8 advanced diffractometer equipped with a Cu seal tube to generate Cu Kα1 radiation (λ = 1.5406 Å) 20 < 2θ < 60° in steps of 0.02°. The collected data were refined by the Rietveld method (Rietveld 1969) using the FULLPROF 05 software, in an attempt to fit the diffraction pattern, in terms of preferred orientation, peak shape, and background. The Thompson-Cox-Hastings pseudo-Voigt function was used to model the peak shape and the preferred orientation. The background was modeled as a polynomial of six refined coefficients. As platform, the Anatase–Rutile template was used. The template was modified by eliminating the rutile phase and varying all the parameters in the next sequence, the scale, background, instrumental, cell and FWHM parameters, preferred orientation, shape, asymmetry, atomic positions, and the overall B-factor (Torres-Martínez et al. 2008). Raman spectroscopy measurements were acquired with LabRAM HR, Horiba Scientific setup equipped with a Nd:YAG laser (λ = 532 nm 80 output power 80 mW). The samples were irradiated with a power of 6 mW over a circular area of 1.5 μm diameter of using a microscope set with a 10× objective.

Photoelectrocatalysis degradation test

Degradation experiments were carried out using an aqueous solution containing Na2SO4 0.05 M pH 3 as electrolyte and 50 mg L−1 of C16H11N2NaO4S (Orange II dye, ALDRICH) as a model dye. In these tests, a vitreous glassy carbon (2 cm2 geometric area) was used as working, the M:TiO2/ITO (M = 5 and 20 wt% of Ni and/or Co; 2 cm2, geometric area) as counter, and a Hg/Hg2SO4 as reference electrode. The experiments were performed under oxygen atmosphere to promote oxygen reduction via two electrons (Esquivel et al. 2009; Peralta-Hernández et al. 2008). The experiments were carried out under potentiostatic conditions applying −0.9 V versus Hg/Hg2SO4 (VoltaLab potentiostat/galvanostat, PGZ 301). For the photo-assisted process, the system was illuminated using a low-pressure mercury lamp (λ = 365 nm, P = 0.75 μ cm−2). UV–Vis measurements were carried out employing a spectrophotometer (Agilent Technologies) to follow the color disappearing of the model dye species (λ = 487 nm, azo bond).

Results and discussion

Characterization of the Co:TiO2 and Ni:TiO2

Scanning electron microscopy images of the modified materials show that the morphology in both semiconductor films (Co, Ni:TiO2) is characterized by aggregates (clusters) distributed homogeneously over the electrode surface (Fig. 1a, b) (Esquivel et al. 2009; Shieh and Chang 2010). EDX analysis, on the other hand, shows the expected presence of the dopant metal (Co or Ni). An approximation of the mole fraction (X i) was estimated using the values obtained from EDX analysis, showing that, as expected, an increase in the weight percentage of dopant in the preparation method, results in a larger relationship between the cation and the titanium species (Table 1). Elemental analysis experiments also showed that in the final material, only a small percentage of the starting quantities of either Ni or Co used is incorporated in the resulting modified substance.

Fig. 1
figure 1

SEM images of 20 wt% samples of a Co:TiO2 and b Ni:TiO2 at ×750, ×1500, and ×5000

Full size image
Table 1 Atomic percentage data and mole fraction (X i), obtained from EDX analysis of the different wt% of Co and Ni doped TiO2 materials
Full size table

To confirm the presence of the dopant, diffuse reflectance spectroscopy measurements were carried out. The Kubelka–Munk function (Ananpattarachai et al. 2009) shows a decrease in the band gap value, an observation that supports the doping of the TiO2 material. The band gap value obtained for the undoped TiO2 was 3.1 eV, which is in good agreement with that reported for commercial TiO2 Degussa P25 (3.2 eV). For the nickel-doped materials, the obtained values show slight changes around 0.1 eV. In this way, the obtained values for the TiO2-modified materials with starting Ni amounts of 5, 20, 40, and 50% were 2.5, 2.4, 2.3, and 2.1 eV, respectively. For the cobalt-modified materials, the resulting band gap values were, 2.7, 2.5, 2, and 1.9 eV for 5, 20, 40, and 50 wt% Co, respectively.

A simple inspection of the obtained XRD patterns for all the materials with different weight percentages of Co and Ni did not allow us to establish the presence of these metals. The diffraction angles detected in 2θ (25.3°, 36.96°, 37.82°, 38.6°, 48.04°, 53.92°, and 55.08°) indicates the presence of the crystalline anatase phase (Fig. 2a, b). From the diffraction intensity, a preferential orientation of the plane (101) can be proposed. In the case of samples with a high-amount of dopant (above 40 wt%), new diffraction signals (between 30° and 36° in 2θ) can be associated to the rutile phase of TiO2. This observation is in agreement with previous studies which reported that an increment in the concentration of a doping cation promotes a transformation from the anatase to the rutile phase (Li et al. 2003; Barakat et al. 2005). The fraction of anatase, X A, in the samples of 40 and 50 wt% was calculated using Eq. 4 (Chen et al. 2008):

$$ X_{\text{A}} = {\frac{1}{{1 + {\frac{{I_{\text{R}} }}{{KI_{\text{A}} }}}}}} , $$
(4)

where K is a constant with a value of 0.79, I A corresponds to the X-ray intensity at the highest peak for anatase, and I R is the intensity at the highest peak for rutile.

Fig. 2
figure 2

XRD patterns at different wt% of a Co and b Ni. Diamond is assigned to the anatase phase and filled circle to the ITO substrate

Full size image

The fraction of anatase for samples of 40 wt% of dopant was 0.888 and 0.975 for nickel and cobalt, respectively. For the samples prepared with 50 wt% of dopant, the results were 0.739 and 0.887 for nickel and cobalt, respectively. This information reflects the effect of the increase of the dopant metal amount in the change form anatase to rutile. The XRD measurements were carried out with the material deposited on the ITO glass.

The XRD diffraction pattern of the clean ITO substrate shows signals at 26.48° and 51.56° in 2θ. For the modified materials on the other hand, these signals are present in all the diffraction patterns and with larger intensities in the deposits corresponding to 20 wt% of Ni and Co. This is probably related to the higher porosity of these films as compared to the rest of the samples (SEM images not shown). As it was previously mentioned, the surface is formed by clusters (Fig. 1), and as the weight percentage of the dopant increases, the deposit becomes more compact and uniform. The nanometric dimension of these materials was evaluated employing the Debye–Scherrer equation (Chen et al. 2008), and for all the samples under study, an average size value of 18 nm was obtained.

Additionally, XRD patterns corresponding to Ni and Co oxides were obtained in order to check the presence of these species in the modified materials. The results show their absence in the doped semiconductor films. At this point, it can be stated that the cobalt or nickel is present in the deposit but it was not known if it is inside the crystalline structure substituting the titanium(IV) or just present as an interstitial dopant of the anatase structure. Induced-coupled plasma (ICP) measurements (<0.006 mg L−1), cyclic voltammetry, and polarography showed no presence of nickel or cobalt in solution after electro and photoelectrocatalysis experiments. This fact suggests that the metal is protected inside the TiO2 structure even in acidic media [the Co and Ni are not stable in acidic mediums (Yang et al. 2009)] or in the presence of H2O2.

Despite that the Rietveld refinement method is mainly used for powder samples, we decided to use it as a first approach for the electrophoretic deposit of M:TiO2 in order to propose the possible location of Co2+ or Ni2+ within the structure of the TiO2, (in an interstitial position or as substitution of the Ti4+). In this way, fitting the experimental XRD patterns was carried out assuming that the metal is actually inside the crystalline structure as a substitutional dopant (Li et al. 2005; Djerdj et al. 2006). In Fig. 3a and b, the Rietveld fittings for the samples prepared from 20 wt% of Co and Ni are presented. In these figures, the observed (○) and the calculated patterns (–) are shown along with the difference between them (residual (∆)) and the Bragg positions (∣).The resulting refinement parameters are listed in Table 2a and b. Inspection of this data shows that the Ni or the Co is inserted at the same position of the Ti4+ (0, 0.25, and 0.375, x, y, and z) in the anatase structure. It is also interesting to note that the fitting parameter χ2 is in some cases slightly above 1.3. This difference is probably due to the fact that the collection data were obtained from a diffraction pattern of a deposited film instead of a powder; which can increase the statistical error (Djerdj et al. 2006). When the Ni and/or Co oxide phases were considered on the refinement, no convergence of the method was achieved.

Fig. 3
figure 3

Rietveld refinement plot for the 20 wt% samples of a Co:TiO2 and b Ni:TiO2. Observed (○), calculated (–), difference between calculated and observed patterns (Δ) and Bragg positions (|)

Full size image
Table 2 Rietveld refinement parameters for (a) Co and (b) Ni-doped TiO2
Full size table

In good agreement with the data reported for the undoped anatase phase (Xu et al. 1999), Raman spectroscopy measurements of the unmodified TiO2 film showed five typical signals (Figs. 4a, 5a), 146(Eg), 199(Eg), 400(B1g), 520(A1g), and 643(Eg) cm−1. When the TiO2 was modified with different amounts of Ni, a slight displacement of the Eg vibration signals (150, 201, and 643 cm−1) was obtained. In the case of Co, a larger displacement was observed due to the lower crystallinity of the relevant vibrations Eg, B1g, and A1g (157, 206, 402, 519, and 634 cm−1). In addition, it is important to note that for weight percentages larger than 40 wt%, it was possible to observe a new signal located at 694.258 cm−1 for Co: TiO2 (Fig. 4b) that can be attributed to the A1g mode of Co3O4 (Hadjiev et al. 1988). Similarly, for the Ni:TiO2 materials, this peak appeared at 706.622 cm−1 (Fig. 5b). In this case, the signal could be attributed to NiO according to the Raman Spectra Database of Minerals and Inorganic Materials (RASMIN). These signals are not seen by XRD due to the lack of crystallinity of the relevant oxides and to the detection limit of the equipment. This observation suggests that the TiO2 structure allows a maximum amount of Ti4+ substitution by the dopant Ni2+ or Co2+. The remaining dopant cations are anchored on the TiO2 surface as another phase, such as Co3O4 and/or NiO.

Fig. 4
figure 4

a Raman spectra corresponding to Co:TiO2 samples with different wt% of the cation and b zoom in the 300–800 cm−1 region to show the presence of Co3O4 (filled diamond)

Full size image
Fig. 5
figure 5

a Raman spectra corresponding to Ni:TiO2 samples with different wt% of the cation and b zoom in the 300–800 cm−1 region to show the presence of NiO (filled diamond)

Full size image

According to all the physico-chemical characterization discussed, the nickel and cobalt species are part of the TiO2 lattice substituting some of the Ti4+ centers. In this way, as the dopant material wt% increases, some fraction of this species grows as an oxide outside of the TiO2 lattice. Additional support comes from the fact that the phase change from anatase to rutile is related to increasing amounts of nickel/or cobalt in the synthetic reaction mixture.

As it has been reported, the highest photocatalytic activity in degradation tests is presented by the anatase phase of the TiO2 (Yanga et al. 2010; Elsellami et al. 2010; Rauf and Ashraf 2009), and a mixture of phases (anatasa-rutile) should then be characterized by a smaller photocatalytic performance. In this context, it was found convenient to carry out some preliminary experiments on the photoelectrocatalytic degradation of a model dye (azo Orange II) in aqueous solution using the 5 and 20 wt% of the Co and Ni doped materials as photoanodes.

Degradation test using the Co:TiO2 and Ni:TiO2 supported on ITO as photoanodes

In the photoelectrochemical experiments for both modified materials (Co and Ni), the color in solution was totally removed after 15 min of running the test (Fig. 6). For the non-modified TiO2, the photoelectrocatalytic activity was substantially lower since the color removal percentage was barely above 50% after 20 min of reaction time. This color removal percentage was achieved with the two modified materials (20 wt% Co and Ni) after 5 min. For 5 wt% of Co and Ni:TiO2 on the other hand, the same level of color removal was achieved in 10 and 40 min, respectively. These preliminary experiments suggest that the presence of Ni and Co within the anatase lattice increases the effectiveness of the electrophotocatalytic process as evidenced by a faster color removal rate of the dye Orange II when it is compared to the undoped TiO2. This observation not only suggests a potential application of these modified semiconductor films as photoanodes in water treatment reactors, but also supports the interpretation of the location of the dopant metals within the semiconductor structure.

Fig. 6
figure 6

Absorbance relationships (color removal of orange dye) using TiO2 and M:TiO2 materials modified with different wt% of the metallic doping agent (5 and 20 wt% of Co and Ni). [OG II] 50 mg L−1; electrolyte pH 3; 0.05 M Na2SO4 solution; degradation time 60 min; UV lamp: λ = 365 nm, P = 0.75 μW/cm2; E = −0.9 V versus Hg/Hg2SO4

Full size image

Conclusion

It was possible to modify TiO2, obtained from metallic titanium, with Co and Ni. Also, it was possible to obtain a deposit of the modified materials on a conductive surface by the electrophoretic method. Using the Rietveld refinement, it can be proposed that the Co and Ni are located inside the TiO2 crystalline anatase structure, substituting some of the titanium (Ti4+) centers. Raman spectroscopy experiments, on the other hand, showed the presence of the oxides Co3O4 and NiO when the starting amount of doping cation was above 40 wt%. In this way, it is possible to suggest that there is some point where Ti4+ ions cannot be substituted by Ni or Co within the lattice. Since in preliminary photoelectrocatalytic tests, a better performance was observed with these modified materials as compared to the case of undoped TiO2, this study also suggests that TiO2 doped with different amounts of Co and Ni can be used in films that constitute a new promising type of photoanodes.