Hydrogen storage in Ti, V and their oxides-based thin films

We have investigated the hydrogen storage ability and the effect of hydrogenation on structure and physical properties of Ti/V and their oxides-based thin films. A series of Ti–TiO2 and VOx–TiO2 thin films with different layer structures, geometries and thicknesses have been prepared by the sputtering technique on different (Si(111), SiO2, C) substrates. For the Ti–TiO2–Ti films up to 50 at.% of hydrogen can be stored in the Ti layers, while the hydrogen can penetrate without accumulation through the TiO2 layer. A large hydrogen storage was also found in some V2O5–TiO2 films. Hydrogen could also remove the preferential orientation in the Ti films and induce a transition of V2O5 to VO2 in the films.


Introduction
Hydrogen has been generally accepted as a fuel in the future. There is currently a lot of hope in hydrogen-based energy. The construction of hydrogen-based energy systems brings up new issues, such as the interaction of hydrogen and matter in the solid state, the search for hydrogen storage materials etc This work is a review of our recent work of investigations of the hydrogen storage as well as the effect of hydrogenation on structure and physical properties of Ti and TiO 2 -based thin films. Some new data obtained for V and VO x -based thin films will be also presented.
An increased interest in TiO 2 has been recently focused on the development of its nanostructured forms (nanotubes, nanorods, etc) for renewable energy sources and hydrogen economy. For example, TiO 2 nanotubes grown on Ti substrates by anodization could be used for hydrogen storage [11]. Metal-insulator-metal (MIM) structures such as Ti/ TiO 2 /Pt were proposed for resistance random access memories (ReRAM) [12,13]. Light induced metal-semiconductor reversible transition at room temperature has been demonstrated for TiO x nanoparticles making it a promising material for high density storage media [14].
The vanadium oxide family (V 2 O 5 , V 2 O 3 , VO 2 ) is among the most attractive class of smart materials exhibiting semiconductor-metal transition. Vanadium pent-oxide (V 2 O 5 ), the most stable vanadium oxide having a wide band gap and being an n-type semiconductor material, has been used especially as optical and electrical devices, e.g. V 2 O 5 thin films can be applied in electrochromic and electrochemical devices [15] and microscale batteries [16]. A large interest is focusing on investigations of TiO 2 -V 2 O 5 thin films to gain the optimal electrochromic properties [17] regarding their potential applications for electrochromic smart windows. The interaction of V 2 O 5 with TiO 2 would strongly affect the structural properties and reactivity of V 2 O 5 ; in particular, it leads to V 2 O 5 dispersion [18,19]. We notice here that due to oxygen reduction by using e.g. hydrogen, V 2 O 5 can be converted to vanadium dioxide VO 2 and a complex mixture such as V 4 O 7 and V 5 O 9 before V 2 O 3 is reached. The promoting effect of a titanium oxide support on the catalytic properties of vanadium oxide is widely recognized [20]. Investigations of thermochromic properties have revealed that VO 2 -TiO 2 multilayers have a higher luminous transmittance than that of single VO 2 film and could yield a large change of solar transmittance at both temperatures below and above semiconductor-metal transition temperature (∼333 K) of VO 2 [21].
Introduction of hydrogen into the crystal lattice in general leads to a modification of both crystal and electronic structure. Hydrogen absorption brings a relatively small perturbation to the system (e.g. the lattice expansion and the hydrogen bonding with other atoms in the lattice). It, however, often implies some new and interesting features in the systems. Thin films and multilayers often play an important role in the improvement of hydrogen absorption rate as well as the chemical and crystal structure stability. On the other hand, the atom mixing, diffusion across the interfaces and precipitation of nanoparticles may also affect the hydrogen uptake-release cycling as well as the thermodynamic properties of the films.
We aim at characterization of the film structure and properties of the TiO 2 -based thin films, in particular the interlayer properties and the influence of hydrogen intake on the microstructure and electronic structure of the films. Numerous TiO 2 -based thin films, consisting of TiO 2 , Ti and VO x layers with a different layer structure (e.g. with single-, bi-and tri-layer structure of different layer sequences), have been prepared by means of sputtering technique and investigated. Selected films subjected for hydrogenation were then investigated focusing on film stability, hydrogen uptake and hydrogen storage under different conditions. Details of the experiments and the main outcome of our study are presented in the next sections.

Experimentals
Thin films consisted of Ti, TiO 2 and VO x layers with different layer sequences and different layer-thicknesses have been deposited by means of magnetron dc pulse sputtering system [22] on Si(111), silica SiO 2 and C-foil substrates. Our analysis of all investigated layers indicates the presence of TiO 2 rutile, while several vanadium oxide forms (V 2 O 5 , VO 2 …) can exist. For the selected film, an additional Pd layer was deposited on the film surface by MBE technique.
Selected films with chosen film-structure have been subjected to hydrogenation at atmospheric atmosphere (1 bar) and at 300°C (so-called hydrogen charging) and/or at high hydrogen pressure up to 102 bar and at room temperature (RT) with different times. For the VO x -TiO 2 series, some samples were charged with hydrogen at 1 bar only once, while some were charged twice. Each charging was for 3 h.
For investigating the possibility of enhancement of hydrogen absorption in the thin film system, an additional palladium layer was deposited on a chosen film by molecular beam epitaxy (MBE) technique.
The film chemical composition, depth profile, layer thickness and structure were determined by combined analysis of x-ray diffraction (XRD), x-ray reflectometry (XRR), Rutherford back-scattering (RBS) and optical spectrophotometry. In order to compare the layer-thickness determination from three independent complementary methods, we also evaluated the layer thickness in nm from RBS data. More details of RBS analysis are described elsewhere [23,24].
The hydrogen profile was determined by means of a secondary ion mass spectroscopy (SIMS) and N-15 Nuclear Reaction Analysis ( 15 N-NRA method). SIMS was carried out by using Cs + primary ions recording positive secondary ions by a CAMECA ims 5f equipment. The reaction 15 N + 1 H → 12 C + α + γ (4.965 MeV) at a resonance energy of 6.417 MeV was used for the 15 N-NRA method to obtain the results. For data evaluation the computer code SRIM was used.

Hydrogen storage in Ti-TiO 2 films
Hydrogen storage in these systems have been thoroughly investigated and reported in our previous publications [25][26][27]. Our investigations revealed that: I. On the as-deposited films of Ti-TiO 2 system: 1. Single Ti nucleates and grows as a compact layer on the well-defined (111) plane of Si wafer-Ti/Si(111). Interdiffusion was not found at the Ti/Si interface, i.e. a sharp interface was always obtained. This Ti film exhibits a strong preferred orientation with (00.1) plane parallel to the substrate, 2. A small amount of interdiffussion was found at the Ti-TiO 2 interface of the bi-layer film and at both Ti-TiO 2 and TiO 2 -Ti interfaces of the tri-layer film. The intermediate TiO 2 layer exhibits columnar structure and there is some intergrowth between the Ti and TiO 2 layers. The interdiffusion is attributed to Ti diffusion along the channels formed between the TiO 2 columns, 3. More precise determination of layer thickness and element concentration, in particular the oxygen concentration, can be obtained for films deposited on C-foils. However, in this case a strong carbon diffusion (up to 10 at.%) into the film was observed. If the film is thin (the layer thickness <30 nm), the carbon can be found even in the surface layer due to carbon segregation from the substrates, II. On hydrogen-charged Ti-TiO 2 films: 4. High hydrogen concentration (storage), up to a value of over 40-50 at.%, was obtained for the top Tilayer, 5. Palladium could act as a good catalyst for hydrogen diffusion in the Ti-TiO2-Ti films. Without covering the film surface by palladium, the hydrogen concentration in the bottom Ti layer has reached only 15 at.%, whereas it increases up to 40 at.% when the film was covered by palladium, 6. Hydrogen could be moved through TiO2 layer without any accumulation there, 7. The preferential orientation in the Ti films was destroyed/disappeared by hydrogen charging under high pressures (pH2 = 100 bar), 8. Large swelling effect was observed for the thick Ti layer (>240 nm) after hydrogen charging at 100 bar. The enhanced hydrogen concentration (enhanced storage) leads to an increase of the film thickness up to 150% of its original value.
The hydrogen profile determined by 15 N-NRA for the Ti/ TiO 2 /Ti/Si(111) films revealing the large hydrogen storage up to 40-50 at.% is shown in figure 1. More detailed results can be found in our previous publication [26].

Hydrogen storage in VO x -TiO 2 films
Prior to the hydrogenation experiments, we have studied thoroughly the properties of layered structures of numerous VO x -TiO 2 films with different film-geometry and thickness. The measured and simulated RBS spectra for the 5 films deposited on silica SiO 2 substrates are shown in figure 2. Estimated layer composition and thickness for each layer in the films are given in table 1.
The RBS spectrum of V 2 O 5 /SiO 2 film (VS1) is characterized by a large V-peak at energy around 1250 keV from the film, a steep Si-and O edge from the substrate, respectively, at 960 keV and 616 keV. Oxygen is present both in the film and in the substrate, thus no separated oxygen signal from the film was observed; its presence implies some increase of the signal on the left-hand side of the O-edge (below 616 keV) in the RBS spectra. The film structure can be described as follows: 1) the upmost surface layer of the film indeed consists of the most stable and common vanadium oxide V 2 O 5 with a thickness of 13 nm (denoted as layer 1 of sample VS1 in table 1), 2) beneath the V 2 O 5 layer is a thick layer with a composition of V 2 O 5−x (layer 2) with x increasing from 0 to 3.0 and thickness of 51 nm, 3) beneath the V 2 O 5−x layer is the stoichiometric VO 2 layer with a thickness of 15 nm (layer 3) and 4) an interface layer consisted of a mixture of VO 2 + SiO 2 was formed as a consequence of V diffusion deeply into the SiO 2 substrate (revealed by the non-zero background between the V-peak and Si-edge). A more detail analysis of the mixture layer V 2 O 5−x (layer 2, thickness 51 nm) indicates that it consists of two sub-layers, the first one with a thickness of 24 nm beneath the surface V 2 O 5 layer 1) is as a mixture of V 2 O 5 , V 2 O 7 (i.e. x = 1.5) and V 5 O 9 (x = 1.4), while the second one (above the VO 2 layer 3) with a thickness of 27 nm is a mixture of V 2 O 3 layer (x = 2.0) and VO (x = 3.0).
Our results indicate that during film deposition, first the VO 2 layer was formed on SiO 2 substrate (with an oxygen content of 33.3%). With increasing deposition time, the 10 at cm , 15 2 while for layer (2) in the sample, VS2 consisted of 85% TiO 2 , 9% VO 2 and 6% SiO 2 0.09 0.06 ) 10 at cm , 15 2 where m, n, p are the corresponding conversion coefficients estimated using the bulk density of different oxides (e.g. ρ (TiO 2 ) = 4.23 g cm −3 , ρ (VO 2 ) = 4.57 g cm −3 , ρ (V 2 O 5 ) = 3.36 g cm −3 ). In most cases, the thickness could be estimated with a good accuracy, since the layer either consists of only stoichiometric oxide (such as TiO 2 , V 2 O 5 , VO 2 ) or the mixture of oxides of the same type (such as (TiO 2 + VO 2 ) or (VO 2 + SiO 2 ) mixture). For such a mixture, since the oxygen content is the same (66.7%), the percentage of different oxides can be easily estimated based on the ratio  table 1). The curves were normalized and shifted for clarity. Table 1. Chemical composition of TiO 2 -VO x films deposited on silica SiO 2 substrates after deposition (as-deposited films): V 2 O 5 /SiO 2 (denoted as sample VS1), V 2 O 5 /TiO 2 /SiO 2 (VS2), TiO 2 /V 2 O 5 /SiO 2 (VS3 and VS4, with the same layer geometry but different thickness) and TiO 2 + V/SiO 2 (VS5). The layer-thickness of each layer (d (nm)), the percentage (%) of each type of oxides (TiO 2 , V 2 O 5 , VO 2 ) presented in the film are estimated using SIMNRA and the mass density of the bulk (see text). Layer (1) denotes the surface layer. The layer (2) of sample S1 consists of V 2 O 5−x with x = 0−3.0 (see text). . In this case, it is more difficult to estimate the exact ratio of different oxides based on only the ratio of M contents, since the O content is different for each oxide. We notice here that besides uncertainty in thickness conversion, the ambiguity in determination of the layer thickness is also related to the layer quality itself. Namely, the sputter deposited films may have some porosities or defects and thus the mass density of the film is certainly different from that of the bulk. The RBS spectrum of V 2 O 5 /TiO 2 /SiO 2 film (VS2) is characterized by a large V-peak at an energy around 1250 keV and a large Ti-peak at an energy around 1200 keV from the film, a steep Si-and O edge from the substrate (at a lower energy than that for the VS1 film, since the film thickness is larger). The thickness of the V 2 O 5 and TiO 2 layer are quite similar and so are the V-and Ti-content in the film, thus a clear minimum between the V-and T-peak was observed. The TiO 2 was first deposited on the SiO 2 substrate and then the V 2 O 5 layer followed in this case. No stoichiometric TiO 2 was found. Instead, a mixed layer (TiO 2 + VO 2 + SiO 2 ) with a thickness of 54 was formed (layer 2, sample VS2) as a consequence from some Si diffusion (6% SiO 2 ) and V-diffusion (9% VO 2 ) into the TiO 2 film. However, the surface layer with a thickness of 68 nm consists of only stoichiometric V 2 O 5 .
In the case of two TiO 2 /V 2 O 5 /SiO 2 films (VS3 and VS4), i.e. deposition sequence is first the vanadium and then the titanium oxide, our analysis reveals that each film consists of only stoichiometric V 2 O 5 and TiO 2 layer. For the VS3 film, the thickness of TiO 2 (64 nm) and V 2 O 5 layer (53 nm) is in the same order of magnitude. Besides, the film is quite thin. Thus, the V-and Ti-signal was combined into one large peak at around 1200 keV in the RBS spectrum. For the VS4 film, the thickness of TiO 2 layer is estimated to be 113 nm, while it equals 71 nm for V 2 O 5 layer (see table 2). Both layers are thicker than those of VS3. Besides, the thickness of TiO 2 film is about 1.5 times larger than that of the V 2 O 5 one, i.e. the Ti content in the film is much larger. This leads to the wide shoulder (V-signal) and the large peak (Ti signal) in the RBS spectra. In the case of TiO 2 +V/SiO 2 film (VS5), the V content is estimated to be 1% for the entire film, i.e. the film composition is 97% TiO 2 + 3% VO 2 . Since the film is thick (193 nm) and the V-content is small, only a broad peak was observed in the RBS spectrum.
Estimated layer composition and thickness of VO x -TiO 2 films after each charging with hydrogen are given in table 2. We focus on analyzing the hydrogen charging results on sample VS2 (with V 2 O 5 as the surface layer) and VS4 (with TiO 2 as the surface layer). These films were charged by hydrogen twice denoted as H(1) and H(2). The increased thickness of the film (%) due to hydrogenation was estimated for the total film thickness (D (nm)) with respect to that of asdeposited film (=(D(after)-D (before))/D (before charging)). As an example, a comparison of the measured and simulated RBS spectra for the V 2 O 5 /TiO 2 /SiO 2 film (VS2) before and after hydrogen charging are shown in figure 3. The Ti-peak Table 2. Effect of hydrogenation on selected films: V 2 O 5 /TiO 2 /SiO 2 (VS2), TiO 2 /V 2 O 5 /SiO 2 (VS4). They were charged by hydrogen twice (denoted, respectively, by H(1) and (H(2)), each charging was at pressure of 1 bar, at temperature of 300°C and for 3 h. The layer-thickness of each layer (d (nm)), the percentage (%) of each type of oxides (TiO 2 , V 2 O 5 , VO 2 ) presented in the film were estimated using SIMNRA and the mass density of the bulk (see text). Layer (1) denotes the surface layer. Due to enhanced Si diffusion from the SiO 2 substrate into the film, the mixed TiO 2 -VO 2 -SiO 2 layer was formed at the film-substrate interface in all cases. The increased thickness of the film (%) due to hydrogenation was estimated for the total film thickness (D (nm)) with respect to that of as-deposited film.
and V-peak with almost equal intensity before hydrogen charging was observed. The effect of hydrogen charging on this film is revealed by a lowering of the Ti-peak and a widening of this peak at the left hand side. It is caused by a decrease of the percentage of TiO 2 in layer 2 as well as the appearance of an extra mixed layer in the interface (see table 2). The total film thickness is increased, respectively, by 7% and 15% after the first and second hydrogen charging. The thickness change is large enough in this case, which can be seen in the RBS spectrum. No visible hydrogen effect can be seen in RBS spectra from hydrogen charging for the other samples (VS1, VS3, VS4 and VS5). A comparison of the measured and simulated RBS spectra for sample VS4 before and after hydrogen charging is shown in figure 4. The RBS spectra before and after hydrogen charging are similar. The effect from hydrogen charging is mostly revealed by the change of the layer content and layer thickness, but the change in the total thickness is in the range of e.g. 2-3% for VS3 and VS4 film.
We concentrate on analyzing the most visible effect from hydrogen charging, i.e. on the V 2 O 5 /TiO 2 /SiO 2 film (VS2). The hydrogen charging leads to a decrease of the layer thickness of the stoichiometric V 2 O 5 layer (layer 1) from 68 nm to 62 nm, while that of the mixed TiO 2 + VO 2 +SiO 2 layer (layer 2) increases from 54 nm to 56 nm after the first charging. The TiO 2 percentage in layer 2 is only 62.5%, much lower than that before charging (85%). The VO 2 percentage increases from 9% to 26.5%. Besides, an extra layer with the same mixture as that of the layer 2 but with a lower percentage of TiO 2 (19%) and VO 2 (19%) and a higher percentage of SiO 2 (62%) does appear (with a thickness of 13 nm). It can be explained as an enhancement of both Ti diffusion into the SiO 2 substrate and Si diffusion out from substrate into the film. We notice here that the film is heated up to 300°C for 3 h upon hydrogen charging. It certainly promotes such diffusion and as a consequence increases the thickness of the interface layers. Indeed the second hydrogen charging induces a larger enhancement; the percentage of SiO 2 in the interface layers is largely enhanced and reached even 75%. Our results clearly show the transition from V 2 O 5 to VO 2 , or in other words, a reduction of V 2 O 5 and increase of VO 2 due to hydrogen charging.
We did not construct the depth profile from the RBS data, i.e. the concentration of each element as a function of the film thickness, just because the relative change between different layers is very small. For a clear demonstration of the hydrogen effect, we construct the film diagram, where the layer thickness of different layers is drawn proportionally with respect to the values given in table 2. Different compositions of different oxides in the layers are presented by using different colours. The film diagram was shown for VS2, VS3 and VS4, respectively, in figures 5 and 6, since the hydrogen charging leads to a change of layer thickness (up to 15%) and composition in these cases by hydrogen.
The obtained results on VO x -TiO 2 films revealed that: (1) Stoichiometric V 2 O 5 and TiO 2 layers were obtained if the deposition sequence was first the vanadium and then the titanium oxide, (2) The V 2 O 5 reduction upon hydrogen charging, i.e. the V 2 O 5 -VO 2 transition, was always observed, (3) In the case when the V 2 O 5 layer is on the surface, the hydrogen charging effect is much enhanced, indicated by a large increase of the total film thickness (up to 15% Figure 3. Random RBS (markers) and SIMNRA (lines) simulated spectra for V 2 O 5 /TiO 2 /SiO 2 (VS2) before (as-deposited) and after two times of hydrogen charging (denoted, respectively, as H(1) and H (2) ). RBS experiments were performed with the incident He + ion energy of 1.7 MeV and the backscattering angle of 171°. The curves were normalized and shifted for clarity. The hydrogen charging was seen by e.g. the relative change in the peak-intensity of Ti-and Vsignal.

Concluding remarks
The most important finding of our investigations is that hydrogen can be stored largely in the Ti layer (with hydrogen content to 50%) in thin films of Ti-TiO 2 system and that palladium could act as a good catalyst for hydrogen diffusion into the films. A large hydrogen absorption can be obtained in the thin films of VO x -TiO 2 system if the surface layer is the V 2 O 5 layer. Besides, the introduction of hydrogen could also remove the Ti preferential orientation and/or induce a V 2 O 5 -VO 2 transition in the films. The dashed line reveals the film thickness after two times of hydrogenation with respect to that of the as-deposited film (thin solid line). The estimated total film thickness before and after hydrogenation is also given. SiO 2 ) in the layers. The dashed line reveals the film thickness after two times of hydrogenation with respect to that of the as-deposited film (thin solid line). The estimated total film thickness before and after hydrogenation is also given.