3.1. Effect of Hydrogenation
The N
2 adsorption–desorption analyses of the neat and hydrogenated SiTi material reveals similar properties. Both samples display a similar type II isotherm and H3 hysteresis loop (
Figure S1, Supporting Information). While this is often associated with macroporous materials, it has also been reported to result from mesopores forming between particle aggregates [
28]. In the case of FSP-synthesized metal oxides, the formation of sinter bridges in the particle agglomerates affords this mesoporosity [
29,
30]. This is confirmed by the broad pore size distributions, displayed in
Figure S2. SiTi and SiTi(H) display similar BET surface areas (98.7 and 96.4 m
2/g, respectively), suggesting that the hydrogenation treatment does not impact morphology or particle size.
A colour change from white to a pale blue was observed in the SiTi(H) sample after hydrogenation. This is attributed to the reduction of Ti
4+ to Ti
3+ [
12,
31], generating oxygen vacancies. The blue colour of SiTi(H) remained constant after hydrogenation, suggesting that the addition of SiO
2 is able to effectively stabilise the Ti
3+ site. The colour change in the material was further investigated with UV-vis analysis (
Figure S3), where the emergence of a broad absorbance band was observed above 400 nm in SiTi(H). The band gap potential remained unchanged (approx. 3.40 eV) after hydrogenation, indicating that the presence of SiO
2 enlarges the bang gap potential above pure anatase (3.20 eV) and rutile (3.02 eV) TiO
2 [
32], but is not affected by the hydrogenation process.
EPR was used to probe the Si/Ti defects generated from hydrogenation, with the results shown in
Figure 1. A sharp peak was observed at a g-value of 2.002 in the neat SiTi sample, which indicates that intrinsic E’ centre defects [
33] are present in the sample. These are suggested to form as a result of the flame spray pyrolysis synthesis [
34]. As flame spray pyrolysis is a rapid process with a very high temperature gradient of cooling, perfect particle rearrangement is not achieved, and E’ centres can be formed. Hydrogenation of the SiTi material did not affect the position of the E’ centre peak, while an increase in peak amplitude for the SiTi(H) suggests that the process may lead to the formation of a greater number of E’ centre defects. A broad peak at a g-value of 1.951 emerges in SiTi(H), with this peak position consistent with the formation of Ti
3+ defects [
19]. This is confirmation that the hydrogenation process is effective in partially reducing Ti
4+ to Ti
3+.
The presence of Ti
3+ defects after hydrogenation was confirmed in the Ti2p XPS spectra (
Figure 2a), where the binding energy of the Ti2p
3/2 and Ti2p
1/2 peaks (observed at 459.2 eV and 464.9 eV, respectively) remained constant, but two additional peaks at 458.3 eV and 464.2 eV are observed in SiTi(H). Analysis of the separate peaks in the Ti2p XPS spectra allowed for the quantity of Ti
3+ in the SiTi material to be determined. While neat SiTi shows no sign of Ti
3+ content being present, as confirmed by EPR and XPS analyses, hydrogenation resulted in a 3.9% content of Ti
3+ in SiTi(H).
The Si2p spectra contains peaks at 103.5 eV and 102.7 eV, which are attributed to Si
4+2p
1/2 and Si
4+2p
3/2, respectively. These peak positions are situated at a lower binding energy than what has been reported for SiO
2 prepared via flame spray pyrolysis [
35]. In this case, the shift of Si2p peak binding energy can be attributed to strong interactions between Si and Ti which are formed during synthesis. Hydrogenation of the SiTi material does not appear to affect this interaction, as the peak positions of SiTi(H) and SiTi are identical. This is confirmed in the O1s spectra, where the peak position of the O-Si (530.5 eV) and O-Ti (532.4 eV) are unchanged after hydrogenation.
Hydrogenation of the mixed SiTi material at 500 °C was an effective method of reducing TiO
2 or Ti
4+, as evidenced by the change in colour of SiTi(H) to blue and the accompanying broad absorbance above 400 nm in the UV-vis spectra. Furthermore, the generation of Ti
3+ and accompanying oxygen vacancies is confirmed by the EPR and XPS analyses. Despite the change in Ti properties after hydrogenation, the crystallinity of the bulk material remains unchanged, as shown in
Figure S4 (Supporting Information). This, combined with the unchanged relative surface area, shows that hydrogenation is an effective method of targeting changes in the TiO
2 structure of the SiTi material.
3.2. Effect of Plasma Treatment
Plasma treatment offers a method of defect generation that is separate to hydrogenation. It is a low energy method, requiring only 27.6 kJ of energy over the course of the 20 min treatment. Thus, the impact of plasma treatment on the structure and performance of mixed SiO2/TiO2 materials before and after hydrogenation is assessed.
The X-Ray diffraction pattern (
Figure S3) of both the neat SiTi and SiTi(H) sample displays a strong anatase crystal phase. A significantly weaker reflection for the rutile phase was also identified from the small peaks at 27.5°, 36.1°, and 41.4°, which are characteristic of the (110), (101), and (111) planes. The anatase/rutile mixture is consistent with TiO
2 synthesized via FSP [
36]. After plasma treatment for 20 min, no discernible change in peak position or amplitude was observed (as shown in
Figure 3), regardless of whether the sample had previously been hydrogenated. This indicates that the anatase/rutile composition and crystallite size of the SiTi is left unchanged in the bulk sample. This does not rule out changes to the surface of the SiTi materials, as XRD does not have sufficient sensitivity to detect such changes, as shown by both Zhuang et al. [
37] (for a He plasma system) and Kong et al. [
20] (Ar plasma).
To further probe the impact of plasma treatment on the crystal microstructure of SiTi, Raman spectroscopy was carried out on SiTi and SiTi(H) after 0–20 min of plasma treatment (
Figure 4). The peak locations in the Raman spectra are in agreement with anatase TiO
2 [
38,
39]. The sharp peak at a wavenumber of 144 cm
−1, shoulder at 195 cm
−1, and smaller peak at 638 cm
−1 are attributed to E
g symmetry. The peak at 397 cm
−1 corresponds to the B
1g symmetry of anatase, while the peak at 515 cm
−1 could represent the superposition of both the A
1g and B
1g symmetry. Plasma treating the neat SiTi sample results in a slight broadening of the peaks and an increase in the ratio of the B
1g signal at 397 cm
−1 to E
g signal at 630 cm
−1. This represents a decrease in local crystallinity as a result of prolonged plasma treatment. An additional peak emerges after 20 min of plasma treatment as a weak shoulder at approximately 440 cm
−1. This peak position is similar to that reported for the E
g phase of rutile [
40], but the lack of a corresponding A
1g peak at approximately 610 cm
−1 suggests that there may be another explanation. For example, peak shift and broadening has also been observed for TiO
2 materials when a surface oxygen deficiency is induced via reduction, where oxygen migrates from surface/subsurface of the TiO
2 into the bulk [
39]. Additionally, a shift or broadening of the TiO
2 Raman spectra has also been reported when significant internal stress occurs [
41]. This will be further explored with XPS analyses of the SiTi materials.
The effect of plasma treatment on SiTi(H) is less pronounced, as shown in
Figure 4b. Hydrogenation results in slight peak broadening, potentially representing a decrease in crystallinity and structural symmetry as a result of the formation of Ti
3+. A similar observation was made for SiTi(H20), but this trend was partially reversed for SiTi(H10), which was plasma treated for 10 min. Unlike the neat SiTi, prolonged plasma treatment of the hydrogenated samples did not result in the emergence of a new peak. This suggests that the reduction of Ti
4+ to Ti
3+ via hydrogenation stabilises the TiO
2 structure in SiTi(H).
Analysis of the surface of SiTi and SiTi(H) with XPS before and after plasma treatment revealed that even 10 min of plasma treatment is effective in generating Ti
3+ defects, as shown in
Figure 5a. While the neat SiTi material contained only the Ti
4+ species, SiTi(10) and SiTi(20) have a Ti
3+ content of 5.5% and 6.5%, respectively. Similarly, the plasma treatment increased the Ti
3+ content of the hydrogenated SiTi(H) from 3.9% to 6.8% after a 20 min plasma treatment. Despite SiTi(10) and SiTi(20) being shown via XPS to contain significant Ti
3+, no colour change to blue was observed. This is explained by the surface nature of the plasma treatment process and low penetration depth of XPS analyses [
42]. Low temperature DBD plasma treatments have been shown to be effective in modifying the surface of materials [
22,
43], rather than the bulk, indicating that the Ti
3+ content after plasma treatment is likely much higher on the surface of the SiTi materials than in the bulk. As a result, SiTi(10) and SiTi(20) remain white in colour.
The Si2p and O1s XPS spectra (
Figure 5b,c, respectively) give an insight into the impact of the plasma treatment on the O and Si species in the SiTi materials. As already discussed, hydrogenation did not result in a shift in the binding energy of the peaks within the Ti, Si, or O spectra. Likewise, plasma treatment of SiTi for 10 min showed no significant peak shifts, however, increasing the treatment time to 20 min resulted in a peak shift of both the Si2p and O1s spectra for SiTi(20) to higher binding energy. This is evidence that the prolonged plasma treatment impacted the Si and O species near the catalyst surface. The shift to higher binding energy suggests that both Si and O have experienced a loss of electron density. This is the same sample which showed the emergence of an extra peak in the Raman spectra (
Figure 4a), indicating that there may be a surface rearrangement which alters the way that Si and O interact on the catalyst surface. The shift of the Si2p spectra to higher BE represents a move towards pure SiO
2 [
44] and could signify a weakened interaction between the Si and Ti at the catalyst surface.
While plasma treatment of the hydrogenated SiTi(H) for 20 min resulted in an increase in Ti
3+ content, it did not impact the Si2p or O1s spectra, where no peak shifts can be observed. This supports the observations from Raman spectroscopy (
Figure 4b) and is further evidence that hydrogenation goes some way toward stabilising the Si–Ti interaction at the catalyst surface against plasma treatment, relative to the neat SiTi material.
Ti K-edge XAS experiments were used to further understand the structural and chemical effects of hydrogenation and plasma processing on the SiTi. The X-ray absorption near edge spectra (XANES) for plasma treated and hydrogenated samples are shown in
Figure 6 and are sensitive to local electronic structure, oxidation state, and local Ti-symmetry. The pre-edge feature of the Ti K-edge, representing 1 s → 3 d electronic transitions, is particularly sensitive to changes in local Ti symmetry and showcases notable structural changes across all samples. For reference, anatase and rutile are used for comparison purposes, along with SiTi and TiO
2 synthesized via FSP without further chemical modification. Anatase exhibits well characterized pre-edge features at 4969 eV, 4972 eV, and 4974 eV, while rutile pre-edge features are positioned at 4971 eV and 4974 eV, with a small shoulder peak at 4969 eV. The materials synthesized by FSP exhibit very similar pre-edge features to anatase, suggesting the Ti atoms hold an anatase-type symmetry that is minimally influenced by the Si atoms, confirming the XRD and Raman analyses. Similarly, pre-edge feature positions for each hydrogenated SiTi material align with anatase, suggesting hydrogenation does not drastically alter the local Ti symmetry.
Unlike the hydrogenated samples, plasma treatment of the neat SiTi exhibits notable structural differences. SiTi(10) exhibits pre-edge features similar to anatase coupled with an additional shoulder at 4970 eV. The shoulder is likely indicative of the formation of five-fold coordinated Ti [
45] rather than the formation of rutile, given that the lowest edge pre-edge feature is still present. The pre-edge features of SiTi(20) exhibit further perturbations away from the parent materials, where the lowest energy pre-edge feature is more similar to that of rutile, coupled with relative peak intensity changes. Examination of the pre-edge feature overall showcases that plasma treatment of SiTi materials influences the local symmetry of Ti, with longer plasma treatment times leading to more pronounced changes in material structure. This is expected to influence the catalytic properties of these materials.
Further examination of the XANES showcases additional changes in chemical and electronic structure of the SiTi materials with hydrogenation and plasma treatment. Overall, the shape of the XANES is very similar to that of anatase for all SiTi, consistent with our analysis of the pre-edge features above. Modest white line intensity increases are observed for the SiTi(H) and SiTi(H20) materials, while a much larger white line intensity is observed for SiTi(20). Increases in white line intensity indicate that the excited electron can access higher order molecular orbitals in the conduction band, almost always due to local electronic withdrawing effects of nearby atoms and/or a change in oxidation state. SiTi(20) also exhibits a large shift in E0 to higher energy, suggesting a bulk increase in Ti oxidation state. This is in line with the Raman analysis of SiTi(20), where peak broadening was attributed to a potential migration of surface or sub-surface oxygen into the bulk TiO2 material.
The extended X-ray absorption fine structure (EXAFS) spectra, presented in
Figure 7, provides location structural information of the Ti–O and Ti–Ti nearest neighbour distances (NNDs). Note that all distances are not corrected for phase shift and are ~0.4 Å shorter than the actual NNDs. Both anatase and rutile exhibit at Ti–O NND 1.56 Å. Unmodified TiO
2 synthesized by FSP exhibits a similar Ti–O NND, while the SiTi material has a reduced NND of 1.50 Å. For the hydrogenated materials, the Ti–O NNDs are 1.50, 1.55, and 1.52 Å for SiTi(H), SiTi(H10), and SiTi(H20), respectively. Note that these NNDs fall between the unmodified SiTi material and pure anatase, strongly suggesting that hydrogenation is affecting the local structure of the material. For the plasma treated materials, SiTi(10) exhibits a Ti–O NND of 1.51 Å, while a NND of 1.43 Å is observed for the SiTi(20). This large contraction of the Ti–O NND indicates that plasma treatment is having a profound effect on the structure of the catalysts, which in turn influences the catalytic properties.
Plasma induced structural changes are even more severely apparent when comparing longer NNDs corresponding to Ti–Ti distances of 2.32 Å and 3.12 Å in anatase. The starting SiTi FSP material exhibits significantly reduced features at these distances, which is expected when considering the nanoscale size of the particles. Upon plasma treatment, the Ti–Ti NNDs shift in the SiTi(10) sample, while a new feature appears in the SiTi(20) material. This feature has a maximum at 2.51 Å and a shoulder at 2.87 Å, which corresponds to an expansion and contraction of the relative Ti–Ti distance in anatase. The origin of these features cannot be determined at this time but must be due to a structural rearrangement of local Ti clusters due to the increased number of defects present in the material as a result of the prolonged plasma treatment. For the hydrothermally treated materials, the shift to longer NNDs is notable for all materials, indicative of a possible lattice expansion, although all samples retain the profile of an anatase-type structure. This suggests that the extended plasma treatment time for the SiTi material after hydrogenation has a less pronounced effect on Ti bonding, supporting the suggestion that hydrogenation assists in stabilising the TiO2 structure.
Plasma treatment of the mixed SiO2/TiO2 material before and after hydrogenation invokes modifications in the material structure and properties. The surface area and crystallinity of the SiTi materials is unaffected by up to 20 min of plasma treatment. Both the SiTi and SiTi(H) materials are altered by plasma treatment, according to the XPS analyses, with the reduction of Ti4+ to Ti3+ occurring among all samples, although the change brought about by plasma treatment was confined to the catalyst surface. XAS analyses demonstrate that chemical/structural changes occur in the SiTi materials, with more pronounced changes observed in the materials which did not undergo hydrogenation. In particular, there is an increase in the oxidation state (from XANES) and the emergence of new Ti–Ti features in SiTi(20). This same material exhibits a peak shift and broadening in Raman spectroscopy and a peak shift in the Si2p and O1s XPS spectra, which represents a change in the surface Si and O species, which may be related to a significant surface oxygen deficiency or internal stress. Hydrogenating the catalyst prior to conducting plasma treatment appears to be an effective way to limit these changes.
3.3. Photocatalytic Formic Acid Oxidation
The ability of the mixed SiTi catalysts to activate oxygen was examined by monitoring the oxidation of formic acid under UV illumination. The mineralization of carbon and the rate of 50% formic acid oxidation (R
50) were determined (
Figure 8). Neat SiTi achieved a low rate of carbon mineralization, with only approximately 60% of the initial 100 μmol of formic acid oxidised to carbon dioxide after 30 min. As a result, the R
50 of SiTi is 1.79 μmol/min. A 10 min plasma treatment of SiTi resulted in a sharp increase in the rate of carbon mineralization, especially in the first 10 min of activity, with SiTi(10) presenting an R
50 of 3.16 μmol/min. This increase in the R
50 can be attributed to the generation of surface Ti
3+ species during plasma treatment. XPS revealed that the plasma treatment is effective in increasing the number of surface Ti
3+ species, which likely contributes to the increase in activity for SiTi(10) relative to SiTi, as the Ti
3+ was identified in a previous study as an active site for formic acid oxidation [
19]. Furthermore, studies investigating the photocatalytic activity of doped TiO
2 materials have shown that, while Ti
3+ is a key component in oxygen activation, it is stable and remains active across multiple reaction cycles [
46,
47].
Increasing the plasma treatment to 20 min for SiTi(20) then caused a drop in performance, with the R50 of 2.42 μmol/min. The changes in material properties offer an explanation for this decrease in activity between 10 and 20 min of plasma treatment. While the plasma treatment did not affect the bulk TiO2 material, it resulted in significant changes to the catalyst surface. SiTi(20) was found with Raman and XPS to exhibit oxygen deficiencies and a decreased electron density for surface Si and O species. Thus, even though the surface Ti3+ content on SiTi(20) (6.5%) was higher than SiTi(10) (5.5%), the weakened interaction between Si/O and Ti at the catalyst surface outweighed this and decreased the photocatalytic activity of SiTi(20).
Hydrogenation of the mixed SiO2/TiO2 material increased its photocatalytic activity, where SiTi(H) achieves an R50 of 2.66 μmol/min compared to 1.79 μmol/min for neat SiTi. This increase in the rate of formic acid oxidation can be attributed to the reduction of Ti4+ to Ti3+ during the hydrogenation process. Plasma treatment of the hydrogenated sample resulted in a further increase in the rate of formic acid oxidation, with SiTi(H10) and SiTi(H20) achieving 92 and 99% conversion of formic acid in 30 min, respectively, corresponding to R50 values of 3.62 and 4.01 μmol/min. This improvement can be attributed to the greater Ti3+ presence on the surface of SiTi(H10) and SiTi(H20) and the stabilization of the SiTi material as a result of hydrogenation. As a result, the plasma treatment of the hydrogenated SiO2/TiO2 material results in further reduction of Ti4+ species present at or near the catalyst surface without impacting the Si–Ti interaction.