Optimization of VTO50 Synthesis

When using the ‘LO’ method to produce rutile samples with 50% vanadium replacement some melting behaviour was noted during sintering. Several tests were conducted to see if a combination of sintering parameters and preparation methods could be used to ensure phase purity and sample quality. A summary of conditions tested and the related samples are presented in Table 7.3.

Table 7.3. Summary of sintering conditions tested for ‘LO’ optimization of VTO50. Optimum conditions used to produce samples studied in later sections are indicated.

Condition Variations tested

Reactant Mixture Ball milled oxides (optimum) and hand milled oxides. Pre-sintering

Steps PVA added to oxides before pressing, pellets pressed to 10 tonnes pressure (opt.), furnace heating rate 4 – 6 °C/min (opt.) and furnace heating rate 8 – 10 °C/min.

Main Sintering

Temperature 610, 800, 850, 870, 900 (opt.) and 1050 °C (opt.*). Main Sintering

Time 15 minutes, 4, 18, 19, 20 (opt.), 47 and 64 hours. Post-Sintering

Steps Grind and re-sinter at higher temperature, and grind and re-sinter at lower temperature. *samples used in the WDXS and XPS measurements

It was found that no combination of double sintering led to an improved product via the solid state method. The most sensitive parameters were the temperature and duration of the sintering step. In Figure 7.3 a time comparison is shown which suggests 800 °C is too low for the reaction even after 64 hours of sintering. While this temperature did appear to reduce melting, longer times showed little improvement.

Figure 7.3. Comparison of VTO50-LO samples sintered at 800 °C for 20 hours (red) and 64 hours (blue). Rutile peaks are indicated by lines, showing additional peaks associated with V2O5 do not vanish with extra time, and 800 °C in insufficient for synthesis.

Increasing the temperatures, it was found 900 °C and above were appropriate for the reaction (Figure 7.4) though unavoidably resulted in distortion of the pellets. A sintering time of 20 hours gave good phase purity and density at these temperatures. In samples that exhibited the most distortion from melting, an additional phase was sometimes observed in the XRPD. This could be mechanically removed with polishing of the samples, and is thought to be associated with a pooling of vanadium melt.

Figure 7.4. XRPD patterns of VTO50-LO samples sintered for 20 hours at 800°C (red), 850 °C (blue) , 900°C (black) and 1050°C (green), implying sintering at temperatures of ≥ 900 °C for 20 hours is optimal for rutile phase formation.

From this work, it is notable that 50% vanadium could be incorporated into these samples without phase separation. This has not previously been shown to be possible with just reducing

gases252 _{and required VO2 and/or vacuum sealing}249, 259-261, 286 _{reducing conditions. This is an} unprecedented result using a low complexity single-step synthesis method.

From this overall synthetic study, a solid state method was successfully developed to produce samples with 1 – 50% vanadium in the rutile phase. The results indicate the solubility of vanadium into TiO2 exists depends on the oxygen partial pressure, and the limits are different to previous reports. This implies a likely change in vanadium valence states on introduction into the rutile framework. The next section explores the chemical nature of vanadium and how it fits into the rutile structure.

7.2

Chemical and Structural Analysis

Noting samples with the rutile structure had been obtained, the change in the size of the unit cell as vanadium was incorporated was investigated. This gave an idea about the location of the vanadium in the rutile matrix and its valence state. A notable change is observed in the rutile lattice of the samples as determined by XRPD. Each fitted to a P42/mnm rutile unit cell and there was a consistent decreasing trend in both the a and c lattice parameters as more vanadium was added (Figure 7.5). It appears the most change occurs before 5% and the unit cell slowly decreases from then onwards. The overall cell shrinkage implies an average decrease in the octahedral site size in the structure.

Figure 7.5. Lattice parameters obtained from fitting of XRPD patterns of VTO samples to a P42/mnm unit cell. A decreasing cell size is noted on increasing incorporation of vanadium.

The size of the three common vanadium valences are V5+ _{– 68 pm, V}4+ _{– 72 pm and V}3+ _{– 78} pm75_{. The size of Ti}4+ _{in octahedral coordination is typically 74.5 pm}75_{, thus the decrease in} lattice size observed may correlate with the titanium being substituted with the smaller V4+ _or V5+ _{ions. Some simulations have shown that vanadium has a tendency to aggregate}267_{, and} 156

short bonds between vanadium can form286, 418 _{related to V}4+_{. At this point it is expected that} substitutional V4+ _{is likely the majority vanadium state that arises from this sample preparation.} However, it is also evident from these diffraction results that the trend in lattice parameters is non-linear. This is not solid solution behaviour, and suggests different composition and conditions lead to different combinations of defects and structural manifestations. The surface is another possible destination for vanadium, most probably in the highly oxidized V5+ _state. This would not necessary result in a decrease in measured bulk lattice parameters. Thus, one way to interpret the lattice trend between the ‘LO’ and ‘O’ methods might be that vanadium is being introduced substitutionally up to 10% doping (decreasing the lattice parameters), but the O method is not reducing enough to encourage deeper doping beyond this concentration, leading it more surface V5+ _{doping near the solubility limit (plateau in lattice parameters to} 10%). When the method is changed to the more reducing ‘LO’ method, deep incorporation of V4+ _{is resumed (and a further lattice shrinkage is observed from 10 – 50%). To verify what is} happening from this lattice plateau to 50% incorporation in terms of vanadium valence, spectral analyses were conducted.