Andjelika Bjelajac*a,
Rada Petrovićb,
Veljko Djokicb,
Vladimir Matolinc,
Martin Vondračekd,
Kassioge Dembeleef,
Simona Moldovaneg,
Ovidiu Ersene,
Gabriel Socolh,
Ion N. Mihailescuh and
Djordje Janaćkovićb
aUniversity of Belgrade, Innovation Center of Faculty of Technology and Metallurgy, Karnegijeva 4, 11000 Belgrade, Serbia. E-mail: abjelajac@tmf.bg.ac.rs
bUniversity of Belgrade, Faculty of Technology and Metallurgy, Karnegijeva 4, 11000 Belgrade, Serbia
cCharles University in Prague, Faculty of Mathematics and Physics, Department of Surface and Plasma Science, V Holešovičkách 2, 180 00 Prague 8, Czech Republic
dInstitute of Physics of Czech Academy of Sciences, Na Slovance 1999/2, 182 21 Prague 8, Czech Republic
eInstitut de Physique et Chimie des Materiaux de Strasbourg, UMR 7504, CNRS, Université de Strasbourg, 23 rue du Loess, BP 43, 67037 Strasbourg Cedex 02, France
fFritz Haber Institute of the Max Planck Society, Faradayweg 4, 14195 Berlin, Germany
gUniversité de Rouen, INSA Rouen, Groupe de Physique des Matériaux, UMR, CNRS 6634, Avenue de l’Université – BP12, 76801 Saint Etienne du Rouvray, France
hNational Institute for Lasers, Plasma, and Radiation Physics, Lasers Department, “Laser-Surface-Plasma Interactions” Laboratory, PO Box MG-54, RO-77125, Magurele, Ilfov, Romania
First published on 12th October 2018
Anodization of titanium film sputtered on fluorine doped tin oxide (FTO) glass was performed to obtain highly ordered ∼2 μm long and ∼60 nm wide TiO2 nanotubes. The titania films were annealed in ammonia atmosphere to enable the doping with N. The annealing did not affect the nanotubular morphology and the porosity remained open which is a very important requirement for further deposition of CdS quantum dots. The analysis done by transmission electron microscopy (TEM) has shown that the N-doped nanotubes have a smaller interplanar distance as compared to the undoped ones, whose interplanar distance corresponded to anatase phase. This difference was attributed to the N doping and the Sn migration from the substrate, as demonstrated by energy dispersive spectroscopy (EDS) combined with electron energy loss spectroscopy (EELS). The near edge X-ray absorption fine structure (NEXAFS) analysis clearly demonstrated that also the doped TiO2 film has anatase phase. Regarding the chemical composition of the studied samples, the X-ray photoelectron spectroscopy (XPS) and synchrotron radiation photoelectron spectroscopy (SRPES) analyses have shown that N is incorporated both interstitially and substitutionally in the TiO2 lattice, with a decreased contribution of the interstitial after ionic sputtering. The shift of the valence band maximum (VBM) position for the doped TiO2 vs. the undoped TiO2 proved the narrowing of the band gap. The CdS/TiO2 films show bigger VBM shifting that can be attributed to CdS deposit. Comparing the absorption spectra of the bare undoped and doped TiO2 samples, it was noticed that the doping causes a red shift from 397 to 465 nm. Furthermore, the CdS deposition additionally enhances the absorption in the visible range (575 nm for undoped TiO2/CdS and 560 nm for doped TiO2/CdS films).
Another proposed approach to further improve the TiO2 based photoanode in terms of absorbance properties, is by doping with anions, among which N showed to be the most prominent. The doping of TiO2 NTs with N can be done by immersing of the NTs in N containing solutions,11 thermal treatment in NH3 gas atmosphere,12 sputtering in N2 atmosphere,13 high-energetic ion implantation,14 adding urea15 or NH4NO316 in the electrolyte for anodization. The advantage seems to be given to the thermal treatment in NH3 atmosphere since it enables simultaneously both transformation of the anodized TiO2 NTs from amorphous to more a conductive crystalline phase and doping with N by a single step process.
A further step in absorption enhancement of TiO2 based photoanode is the deposition of a sensitizer. The well-known Grätzel cell includes TiO2 sensitized with organic dyes.17 However, that system proved unstable. Thus the use of inorganic quantum dots (QDs) as sensitizer was considered as alternative.18 The most important feature of QDs is the multiple exciton generation (MEG), a phenomenon described as a process of converting the high energy photon into more than one electron.18 In order to provide the efficient electrons transfer from excited QD to TiO2, it is essential to properly align the energy levels of the materials, such is the case of CdS QDs whose conductive band edge is 0.5 eV above the conductive band edge of TiO2.19,20 The effective use of mercaptosilane (MS) as a binding reagent for TiO2 sensitization with CdS QDs was previously reported.21 The role of MS was not just to enable the binding of CdS QDs to TiO2 surface but also to protect colloidal particles from aggregation during the synthesis and aging. When aggregated, QDs lose the quantum confinement effect and consequently the MEG process fails. It is therefore essential to firstly obtain the stable CdS QDs sol and then use it for TiO2 sensitization.
Under this general context, the goal of this work was to boost the absorption properties of TiO2 NTs, both by doping with N as well as by sensitizing with CdS QDs and to study the complementary of the two approaches.
Then, the sputtered titanium films were anodized in ethylene glycol containing 0.3 wt% ammonium fluoride and 2 wt% water. The ethylene glycol based electrolyte was used since it is not aggressive to glass substrate and also provides longer NTs (>2 μm) compared to NTs fabricated using usual HF (aq) electrolyte.24 The Ti/FTO anode was kept 10 mm apart from the platinum cathode. The voltage was set at 60 V and the anodization was carried out until the films became transparent (after ∼10 min). The samples were then well rinsed with water and let to dry in air for 24 h. One half of the samples was annealed in air for 30 min at 450 °C with a heating rate of 8 °C min−1 to induce crystallization of the initially amorphous nanotubes. In that way the undoped TiO2 NTs were obtained. Another half of anodized TiO2 films was annealed in NH3 atmosphere at the same heating and cooling rate to cause the nitrogen doping of the films.25
The sensitization of the TiO2 NTs with CdS QDs was reached by immersing the TiO2 NTs films into CdS colloid for 48 h. The samples were dried for 24 h in air and further characterized.
The following expression of the EMM was used to derive the size of CdS nanoparticles:26,27
(1) |
The surface morphology of the TiO2 NTs before and after deposition of CdS QDs was studied using a Tescan Mira X3 field emission scanning electron microscope (FESEM). The distribution of CdS QDs within the TiO2 nanotubes was analyzed by high resolution TEM (HRTEM) Jeol JEM-2100F and Jeol JEM-2100 operating at 200 kV and the chemical mapping was performed using energy dispersive spectroscopy (EDS) and electron energy loss spectroscopy (EELS). For TEM analysis, the samples were scratched with a diamond needle and the resulting fragments were collected on a grid. The EDS results were provided by analyzing several such obtained fragments. For EELS analysis, the N-doped sample was drowned to epoxy on both sides. Then the etching of the glass substrate was performed with HF until the glass was consumed. The cutting of the remaining film from the opposite side was done using a diamond ultramicrotome knife until the ∼70 nm thin film was obtained and transferred to a carbon grid.
For determination of phase composition of the annealed films, near edge X-ray absorption fine structure (NEXAFS) measurements were performed using synchrotron radiation.
The laboratory X-ray photoelectron spectroscopy (XPS) preliminary analyses (wide scan survey) were performed using 1253.6 eV Mg Kα beam source to check the overall chemical composition of the samples. The high-resolution synchrotron radiation photoelectron spectroscopy (SRPES) enabled the extraction of the N 1s line from the Cd 3d line, expected at 400 eV, in the photoemission spectra of the doped sample with CdS. By varying the incident energy of the beam it was possible to distinguish the valence band position of the reference undoped sample and to compare it with the N-doped sample with and without of CdS. The analyses were performed at the Materials Science Beamline at Elettra synchrotron light source in Trieste.
The diffuse reflectance spectra (DRS) of the undoped and doped TiO2 NTs film formed on FTO glasses and nanocomposites obtained after deposition of CdS QDs were investigated using a Shimadzu 2600 UV-Vis spectrophotometer with an integrating sphere attachment in the wavelength range from 300 to 750 nm. A bare FTO glass was taken as a reference for measuring the baseline.
Taking into account that the band gap of the CdS QDs, , is 4.30 × 10−19 J (2.68 eV), the estimated size of CdS QDs was calculated to be 4.5 nm. This value corresponds to the biggest nanoparticle present in the analysed colloid. If we take into account that MS is on the surface of the nanoparticles, it can be said that the EMM approximation was in accordance with TEM observation since the biggest nanoparticle that was detected was 5 nm.
Fig. 2 presents the SEM micrographs of TiO2 NTs after annealing in air and NH3 with a length of ∼2 μm and ∼60 nm inner diameter. It was observed that the open nanotubular structure is preserved both cases. This is important for the subsequent deposition of CdS QDs.
Fig. 2 SEM micrographs of the anodized TiO2 nanotubes: (a) after the annealing in air, (b) after the annealing in NH3 (length ∼ 2 μm, inner diameter ∼ 60 nm). |
SEM analysis was repeated after 48 hours immersion of TiO2 substrates into CdS colloid. The results are provided in ESI as Fig. S1.† The deposition of CdS QDs was not uniform. There were sporadic areas that cover the nanotubes tops for both undoped and doped TiO2 films. The EDS mapping confirmed that those areas consist more of CdS (Cd:S = 1.07:1.69 at% in average for the undoped sample and 0.57:0.62 at% for the doped sample). However, CdS was also detected within the uncovered film surface (Cd:S = 0.33:0.56 at% in average for the undoped sample and 0.10:0.14 at% for the doped sample). Since S originates both from CdS QDs as well as from MS as a surfactant, it was therefore expected to detect a higher amount of S with regards to Cd (Cd/S relative ratios were all <1). The EDS mapping were recorded from the areas displayed in Fig. S1† and it was used just to confirm the overall presence of Cd and S within the structure. Most of the TiO2 films remain open as they were before deposition without indication of the deposit presence. However, the SEM/EDS characterization gave an insight of the surface composition and morphology only. In order to analyse the structure along the nanotubes depth, the TEM analyses have been performed by using the preparation method described in Experimental part.
The HRTEM micrograph of the undoped TiO2 samples after the 48 h immersion in CdS sol is presented in Fig. 3(a). The nanoparticles are clearly visible within the nanotubular structure and they are ∼5.5 nm in size. It can also be observed that the wall thickness of the nanotube is ∼20 nm. The measured interplanar distance (d = 0.342 nm) of the nanotube's fragment presented in Fig. 3(b) corresponds to (101) plane of the anatase phase of TiO2 (d = 0.352 nm). The 0.010 nm difference could be a consequence of a presence of some other elements in TiO2 structure, for example Sn, as a result of diffusion from FTO support during the annealing in air.
The presence of CdS nanoparticles inside the nanotubular structure of the doped TiO2 film was evident from the micrograph given in Fig. 3(c). The interplanar distance of the nanotube's fragment from Fig. 3(d) is d = 0.224 nm, which is close to the value characteristic for (112) plane of anatase (d = 0.233 nm), but also for the (200) of rutile (d = 0.230 nm). This crystallography difference could be primely due to N incorporation originating from the annealing in NH3. However, it could also be due to the presence of other elements, as noticed for the undoped sample.
By having the advantage of using the synchrotron beamline it was possible to explore the phase composition of TiO2 annealed in ammonia, in relation with the change in crystallography observed in HRTEM. Anatase phase structure of the TiO2 film was evidenced by NEXAFS analysis as depicted in Fig. 4. Ti L2,3 edge presented in Fig. 4(a) and O K edge in Fig. 4(b) were analyzed and compared to literature. For Ti L2,3 edge, the position of six characteristic peaks originating from the splitting of L2 and L3 edges, should be analyzed as explained before.28 However, the differences of the peaks positions between anatase and rutile are negligible in the given spectrum. The distinction could be made only by comparing the intensities of the peaks at 460.5 eV and 461.0 eV. In the case of anatase phase the first one has higher intensity,29–31 as in the spectrum given in Fig. 4(a). The O K edge exhibits two characteristic peaks at 531.4 eV and 534.1 eV, with the separation of 2.7 eV, which matches to the reported value for anatase,28 in contrast to rutile phase that has wider separation (3.0 eV) of the peaks positions. Another observation that confirms the stated phase is the shape of high energy spectrum with two wide shoulders centered at 539.7 eV and 545.6 eV. In the case of amorphous TiO2, there is only one shoulder in this energy range, while in the case of rutile there are three shoulders.28,32
The results of the EDS analyses performed on TEM are summarized in Table 1. Excepting the presence of Ti, O, Cd, S, Si, C and N for the doped sample, there is a significant amount of Sn detected in all analyzed fragments. We advance here the assumption that Sn diffused into TiO2 structure during the annealing,33,34 as it can be the case of Na and Ca35 that are also detected in small amounts. It must be noted that Cu originates from the classical TEM grid used as sample support in the analyses.
Element | Undoped TiO2/CdS | Doped TiO2/CdS |
---|---|---|
C | 52.20 | 60.91 |
N | 0.00 | 1.89 |
O | 21.49 | 11.46 |
Na | 0.18 | 0.60 |
Si | 1.82 | 3.55 |
S | 1.55 | 4.24 |
Ca | 0.08 | 0.00 |
Ti | 15.97 | 3.72 |
Cu | 4.67 | 10.77 |
Cd | 0.67 | 2.16 |
Sn | 1.33 | 0.72 |
According to the Sn/Ti ratio (0.083 for the undoped TiO2/CdS and 0.194 for the doped TiO2/CdS), the incorporation of Sn in the TiO2 structure was more intensive in doped sample. This is in agreement with the high reactivity of SnO2 to NH3, as reported before.36 Actually, SnO2 acts as a sensor for NH3 even at room temperature.37 Therefore, one may assume that during annealing in NH3 at 450 °C, Sn from FTO diffused more into TiO2 structure than during the annealing in air.
Due to a relatively low sensitivity of the EDS technique for the detection of light elements such as N, additional analyses were performed using EELS, a more appropriate spectroscopy method to analyse the presence of N within the doped sample.
In order to access the NTs structure more distinctly, the preparation of the sample was done by slicing the sample perpendicularly to the substrate to ∼70 nm thin film using a diamond ultramicrotome knife as described in Experimental part. That way only the sensitized NTs could be analyzed in cross section without the possible contribution of the substrate. The STEM micrograph of a representative fragment is shown in Fig. 5(a). The NTs decorated with nanoparticles are clearly visible. The EELS spectrum, associated to the image, is displayed in Fig. 5(b). Herefrom one can observe the Ti L2,3 line doublet at 460 eV and O K line at around 535 eV. A further zoom on the spectrum (inset of Fig. 5(b)) has indicated a weak N K edge at 400 eV suggesting the doping with N. Thus, the studied sample still needed a deeper insight for a sure confirmation of N doping in TiO2 structure. The corresponding XPS analyses were therefore performed.
Fig. 5 (a) STEM image of the analyzed fragment, (b) energy loss spectrum of the doped TiO2 nanotubes decorated with CdS (a zoom of N K edge is given in the inset). |
The wide scan XPS spectra as recorded from the undoped and doped TiO2 films are shown in Fig S2.† The analysis of the spectrum in Fig. S2(a)† confirms the presence of TiO2, as well as of a small amount of Sn by detecting the Sn 3d and Sn 3p3/2 lines at 487 eV and at 717 eV, respectively. The spectrum of the doped sample (Fig. S2(b)†) shows a higher amount of Sn. This confirms the assumption that Sn diffuses more from FTO substrate when TiO2 is annealed in NH3. The main difference between the two spectra is the presence of N 1s line at 400 eV in the case of doped sample, quite expected for the specimen synthesized by annealing in NH3 atmosphere. The amount of N within the doped sample was estimated to be ∼1 at% calculated by quantitative analysis taking into account peak area/photoionization cross sections.
The confirmation of N-doping of the sample annealed in NH3 atmosphere, is provided in Fig. 6, where XPS spectra of the N 1s line before and after ionic sputtering are given. The sputtering was applied to clean the surface contaminants and to analyze the N occurrence deeper in the structure. Before sputtering the N 1s line was fitted to one wide peak at 400.3 eV (Fig. 6(a)). It was revealed that after sputtering N 1s line has three contributions at ∼400.9 eV, ∼399.7 eV and ∼397.4 eV (peak 1, 2 and 3 in Fig. 6(b), respectively). The first one at 400.9 eV can originate from chemisorbed N2 molecules on the surface of the TiO2.38 The second contribution at ∼399.7 eV can be attributed to interstitially incorporated N39 and/or NH3 that might be bound to defect sites or trapped in the subsurface region.40 The third contribution at ∼397.4 eV can be interpreted as nitride, meaning that N is incorporated substitutionally in the structure.40
The whole range spectra of the undoped and doped samples with CdS are given in Fig. S3.† Apart from the expected lines for Ti, O, Sn and C, one can notice characteristic lines for Cd, S and Si that confirm the presence of CdS with MS on the surface of TiO2. The signal for S originates from both sulfide from CdS core and from MS shell. The signal of C comes from atmospheric CO2, but also it can be attributed to the (–CH2–) groups originating from MS. In addition, there is Si 3p line detected at 99 eV that is assigned to silane groups present in MS around the CdS nanoparticles. In Fig. S3† in the case of CdS/N-doped TiO2 film, an overlapping exists in the ghost peak of Cd 3d line with N 1s line at ∼400 eV.41 That is why only a high-resolution SRPES analysis provided by synchrotron radiation could resolve this issue.
The SRPES investigation was performed on the doped sample with CdS by monitoring N 1s line before and after 10 min of Ar+ sputtering. From Fig. 7 it can be observed that N 1s line has three main contributions and the correspondent intensities changes after sputtering. The first contribution (peak 1 in Fig. 7) at ∼401.5 eV is decreasing, while the ratio of the second contribution at ∼399.9 eV (peak 2) to the third at ∼397.6 eV (peak 3) remained approximately the same after sputtering. This was expected as the first contribution corresponds to chemisorbed N2 molecules on the surface of the TiO2 and they are easily removed by sputtering.38 The second peak at ∼399.9 eV can be attributed to interstitially incorporated N39 and/or NH3 molecules.40 The third contribution at ∼397.6 eV can be associated to N substitutionally incorporated into TiO2 structure.40
In order to investigate the effect of the doping and deposition on the narrowing of TiO2 band gap, the incident energy of the synchrotron beam was set at 630 eV for all analyzed samples. This approach allows to clearly distinguish the positions of the valence band maximum (VBM). The overall comparison of the VBM is presented in Fig. 8(a). The VBM positions were estimated to be: 3.95 eV, 3.84 eV, 2.81 eV and 2.91 eV for undoped TiO2, doped TiO2, undoped TiO2 with CdS and doped TiO2 with CdS, respectively. The shift of VBM for the doped TiO2 vs. undoped TiO2 can be a consequence of mixing N 2p and O 2p states and/or introduction of N induced donor level formed within the band gap slightly above O 2p valence zone.42 As demonstrated by the results in Fig. 6 and 7, N was incorporated both interstitially and substitutionally. The CdS/TiO2 films show bigger VBM shifting that can be attributed to CdS deposit.43
To assess the effect of doping and CdS sensitization on the absorption properties of TiO2 NTs film, the absorption spectra of the bare undoped and doped TiO2 samples were compared with the corresponding spectra of the samples with CdS (Fig. 8(b)). The absorption edges were read at the intersection of the tangent of the absorption curve and the x axis. For the undoped bare TiO2 film, the absorption edge was estimated to be 397 nm, whereas the doped TiO2 film shows an absorption red shift with the edge at 465 nm. The absorption curves of the CdS sensitized TiO2 films proved that the CdS deposition causes the absorption red shift to 575 nm for undoped TiO2/CdS and 560 nm for doped TiO2/CdS films. The smaller red shift of the doped TiO2/CdS as compared to the undoped TiO2/CdS is in agreement with the VBM positions derived from Fig. 8(a). The corresponding band gaps, Eg, of the films were calculated to be: 3.12 eV, 2.66 eV, 2.15 eV and 2.21 eV for undoped TiO2, doped TiO2, undoped TiO2 with CdS and doped TiO2 with CdS, respectively. Taking into account these values, the positions of the conduction band minimum (CBM) were calculated as:44
CBM = VBM − Eg. | (2) |
The schematic display of energy level positions is given in Fig. 8(c) with the purpose of investigation of the interactions between CdS and TiO2. The Fermi level, Ef, of bare undoped TiO2 is located at 2.39 eV and quasi-Fermi level, , for the undoped TiO2/CdS composite is at 1.74 eV. Following the alignment of the Fermi levels (difference is equal to the difference ), we calculated that Ef(CdS QD) is at 1.08 eV. Having in mind that the Eg of CdS is 2.68 eV, the energy level bands for CdS were 2.42 eV and −0.26 eV, for VBM and CBM respectively. When analyzing the energy level positions of the doped system in the same manner as for the undoped the results lead to the same values for CdS QD energy levels positions. This means that the interaction of CdS and TiO2 remains the same irrespective of the doping. This could be due to the MS presence on the surface of CdS QDs.
Another observation is that the band gap of undoped TiO2 is narrower (3.12 eV) comparing to the value characteristic for anatase (3.2 eV4). We believe it is due to Sn incorporation from FTO substrate as suggested in TEM analysis since there is a 0.01 nm difference in interplanar distance that was measured and that of anatase phase. Also, in the case of the doped sample the CBM is moved compared to the undoped one. That can also be due to the loading of Sn from the support. The annealing in NH3 enhanced the diffusion of Sn and it is assumed that Sn acted as cation dopant creating a donor level in the energy gap of TiO2 resulting in additional narrowing of the band gap.44 This assumption has being further studied in our group.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ra06341a |
This journal is © The Royal Society of Chemistry 2018 |