Chapter 6: Vanadium and Nitrogen Modified Anatase-type Titanium Dioxide Materials
6.2 Chemical and Structural Analysis
6.2.2 Chemical Analysis
6.2.2.1 X-ray Photoelectron Spectroscopy Valence Analysis
Determining the valance state of the cations and presence of dopant anions in these materials is of key importance to understanding their functionality. Increased catalytic ability is often said to be associated with V5+ located at the surface of the particles, in comparison to buried V4+ states215, 219, 226, 404. A good technique to investigate this is XPS. While the average penetration depth of this technique is only 10 nm, this matches well with the experimental particle sizes of these samples. This means that representative dopant types, levels and their valence states can be detected.
Figure 6.4 shows characteristics peaks for V (a), Ti (b) and N (c) in the XPS spectra from each material. An absence of Ti3+ in these samples is noted which contrasts to some references214, 215, 218-220. Two signals for nitrogen were detected which are consistent with other reports which claim bulk N doping214-216, 218, 219. However, the biggest deviation between these materials and other studies, is observed in the V2p3/2 peak.
The V2p3/2 peak of each sample is very broad (Figure 6.4 (a)) suggesting two or more valence states are present in each. This peak was fitted for each sample and the resulted are presented in Table 6.3. The binding energies determined for each peak were assigned to V3+ and V4+ based on previous reports216, 405-407. Generally, as the amount of nitric acid was increased the total amount of vanadium increased, and the proportion of lower valence states also appeared linked to an increase in nitrogen. A numeric summary of each vanadium valence state/signal as a proportion of the cations in the sample, and the nitrogen signals as a proportion of the bulk anion signals is given in Table 6.3. The calculated formulae for the meantime assume both dopants are substitutional and is used simply illustrate relative concentration of elements in the product.
Figure 6.4.XPS spectra of the (a) V2p3/2, (b) Ti2p3/2 and (c) N1s peaks for each VTON sample. A broad, multivalent V peak is seen in each. There appears to be a correlation between the V valence spread and shape of the N1s peak.
Table 6.3. Proportion of vanadium valence states from V2p3/2 peak fitting of VTON materials and the determined composition in comparison to nominal dopant ratios. Fitted binding energies are given in brackets (eV).
Sample Nominal Dopant
Ratio V:Ti:N % V3+ % V4+ Composition from XPS VTON-A* 0.05:1:9 29 (514.95) 71 (516.15) V0.035Ti0.965O1.9460N0.0536 VTON-B1 0.05:1:5 35 (514.74) 65 (516.02) V0.030Ti0.970O1.9740N0.0258 VTON-B2* 0.05:1:10 28 (514.94) 72 (516.07) V0.034Ti0.966O1.9620N0.0380 VTON-B3 0.05:1:20 56 (514.84) 44 (516.13) V0.042Ti0.958O1.9262N0.0738 VTON-B4 0.05:1:40 61 (514.75) 39 (515.93) V0.048Ti0.952O1.9450N0.0550
*Two peak fit
It should be noted that the A and B2 materials have a very similar amount vanadium, and can be fitted with similar proportions of V3+ and V4+. However, these materials do show a higher binding energy shoulder, allowing them to be fit similarly well with three components (Figure 6.5. (a), (c)). In such a fit, only 19% of the total vanadium would be in this higher valence state, which is still lower than anticipated from the (V,N) co-doping simulation studies207-213. This high energy peak would be centred at 516.7 eV in both cases, which is below the typical value quoted for V5+ 399-401, and so it is not clear that this is a true result. The two-component fit is considered the better depiction for this study (Figure 6.5 (b), (d)).
Figure 6.5. Alternate fitting of V2p3/2 peak for sample VTON-A and B2 suggesting a three-component fit might also be valid (a) and (c), but the two-component fit (b) and (d) has more consistent peaks centrings with the literature and is deemed a better model.
A prevalence of the V3+ and V4+ valence states is apparent in these samples, even when V5+ starting materials are used. Every literature report studied in preparation for this work did report or show evidence for more than one valence state in their product, irrespective of the starting valence of the raw material (3 217, 4 216, 218 and 5+ 214, 215, 219) which is attributed to the low energy associated with the reduction of vanadium ions in solution reactions226. However, only one report reviewed for this thesis noted a 3+/4+ vanadium valence prevalence226. They attribute this to bulk doping as opposed to surface doping of vanadium ions which are readily oxidized to V5+.
Other studies have also shown from spectroscopy that V5+ is likely to be at the surface, whereas V4+ is in the bulk24, suggesting vanadium may be doping well into the bulk of the particles. Octahedra in the anatase structure can share four edges with their neighbours instead of the maximum of two in rutile, so V5+ may have less of a preference for substitutional coordination in the bulk of anatase than V4+ ions according to Pauling’s rules300. The anatase structure also has voids in which lower valence ions can fit comfortably, such as V3+ which might go a way to explaining this observation.
The influence of the nitric acid seems particularly apparent in the correlation between the shape of the nitrogen XPS signal (increase in 402 eV edge), and the presence of a large proportion of V3+ (Figure 6.4 (a) and (c) above, samples B1, B3 and B4 in particular). Interestingly, despite this link, there is no clear evidence for vanadium nitride (V3+-N3-) type direct bonding (N1s peak at ~ 398 eV408). A recent publication does suggest that the lower binding energy edge to the V2p3/2 peak (that has been ascribed to V3+ in this work) might be evidence for direct V-N bonding220. However from a consideration of charge compensation within the anatase structure, and absence of corresponding nitride peak, there is not clear evidence for such a defect in this study. The nitrogen signal that is measured in this study reflects other reports206, 214-216, 218, 219 , indicating it has been successfully incorporated, but may not be strictly substitutional, a concept that remains contentious in the literature. From these observations, it seems V3+ and N form correlated defects, but at least one would likely be in non-substitutional geometry.
Overall, there is a tandem doping effect whereby the ratio of V:N in the reaction mixture has a linear relationship with the proportion of vanadium detected in the product. This effect has been noted in other reports of (V,N) co-doping215 and in similar systems205 and has been thought to be related to direct bonding between V5+-N3-, predicted computationally, but the results from this work suggest it is not that straightforward. A prevalence of V4+ is noted all samples. This matches well to the general decrease seen in the structural parameters as V4+ has a smaller radius than Ti4+ in an octahedral environment75. Clusters of V4+ would also result in the
decreasing effect. However the larger V3+ and N3- ions, while present, do not seem to cause a lattice expansion and may suggest they possess more complicated coordination geometries. To summarize, the XPS points to successful doping of vanadium and nitrogen, however the vanadium exists in two reduced states. This suggests the vanadium exists deeply doped in the samples rather than at the surface. The nitrogen appears to help more vanadium be introduced, but there is no clear evidence for the formation of V5+-N3- clusters, rather some substitutional V4+ and some correlated V3+ and N3- with complex geometries. In order to further confirm which species do exist in these samples and isolate their role in functionality, ESR spectroscopy was used.