(1) Powder Synthesis
The Fourier-transform infrared spectrometry (FTIR) absorption spectra for all the HfB2 precursors formed with different Hf:C:B molar ratios and dried at 250°C for 2 h were very similar; one of the spectra across the 4000–400 cm−1 range is shown in Fig. 2. The broad peak at around 3400 cm−1 is attributed to the O–H stretching vibrations associated with the hydroxyl groups in the phenolic resin, boric acid, and the Hf–OH stretching vibrations. The peaks at 418, 1392, 2841, and 2914 cm−1 are attributed to the aliphatic C–H stretching vibrations from the phenolic resin, whereas those at 543, 1099 and 1622 cm−1 represent the aromatic C–H deformation and C=C deformation from the phenolic resin. The peaks at 2330 and 2357 cm−1 are due to C=O stretching arising from the CO2 and C3O2+ groups present in the system. The O–B–O stretching vibrations peaks are present at 449, 520, and 675 cm−1 and these are associated with boric acid present in the system. The peaks at 651 and 748 cm−1 are attributed to the OHfO-asymmetric stretching. The FTIR result confirms that there was no complex formation between the reactants and that at the end of the sol–gel process and subsequent drying step, the precursor powder consisted of H3BO3, Hf–OH, resulting from the hydrolysis of HfCl4, and aromatic and aliphatic chains resulting from the cross-linking of the phenolic resin.
Figure 3 shows the TGA/DSC curves of boric acid and phenolic resin. TG analysis shows that the main weight loss occurred below 320°C for H3BO3, due to the evaporation of the bound water, and below 640°C for phenolic resin, due to the decomposition of the phenolic resin. H3BO3 loses 3 molecules of water at 150°C, 189°C, and 210°C forming metaboric, tetraboric, and pyroboric acid. Further heating above 300°C led to the formation of boron trioxide and the weight stabilizes. The phenolic resin, however, continues to lose weight until 640°C, due to the decomposition resulting in evaporation of physically and chemically bound water. The percentage of the mass remaining at the end of the decomposition process is assumed to be the carbon content of the resin and was used for stoichiometric calculations.
Figure 4 shows the XRD patterns for the HBeC precursor powders heated from 600°C to 1500°C with a 0.1 h dwell in a flowing Ar atmosphere. The results show that monoclinic HfO2 was formed at 650°C. At this temperature, the degradation of phenolic resin to carbon and dehydration of boric acid to boron trioxide will also have been complete, Fig. 3. Therefore, at 650°C the powder mixture consisted of unreacted but intimately mixed B2O3, HfO2, and C. By 1300°C HfB2 peaks formed, indicating the onset formation of HfB2.
Figure 5 shows the indexed XRD patterns of HBC and HBeC heat treated at 1600°C for 2 h. The patterns were indexed according to the JCPDS cards 00-039-1491 and 00-038-1398 for HfC and HfB2, respectively. It can be seen that the pattern corresponding to the HBC precursor, which was made using the stoichiometric mix of the 3 elements, includes significant amounts of HfC impurity3, whereas the HBeC XRD pattern shows only HfB2 peaks. This confirms the need to include excess B in the initial mix to allow for B2O3 losses by vaporization, as observed by Ni et al. The boron loss depended on various factors like humidity, flow rate of the inert gases, and substrate (99.9% alumina boat that contained the powder sample) used. For this reason, the ratio of B:Hf was kept between 3:1 and 3.8:1 instead of 2:1 for subsequent compositions.
The isothermal TGA showed that 25 h was needed at 1300°C to complete the boro/carbothermal reaction and achieve hexagonal HfB2 with no secondary phases, this was confirmed by a heat treatment in tube furnace at the same temperature and dwell time, Fig. 6; only 2 h was required at 1600°C.
As seen from Fig. 7, the products from HBC and HBeC have a very similar particle size of about 1–2 μm indicating that an increase in boron oxide content does not have any effect on the particle size of the end product. The HfB2 powder obtained from HBeC precursor powder was subjected to XPS analysis, Figs. 8(a) and (b). As seen from the spectra, the peaks at 17.7 and 185 eV correspond to the Hf–B bond in HfB2. No obvious peaks corresponding to free boron or HfO2 were detected.[22, 23] The weight percentage of carbon in the same powder as determined by elemental CHN analysis was around 0.19 wt%.
Figure 7. Field-emission gun scanning electron microscopy micrographs showing the particle morphology for (a) HBeC, (b) HBC, heat treated at 1600°C for 2 h (c) HBeC heat treated at 1300°C for 25 h.
Download figure to PowerPoint
Reducing the synthesis temperature from 1600°C to 1300°C and increasing the dwell time from 2 to 25 h for the HBeC precursor powder had the effect of slightly decreasing the particle size from 1–2 μm to 0.25–2 μm, however, the latter heat-treatment conditions also yielded a high proportion of rod shaped particles, Fig. 9(a). It is believed that the particles grow into rods due to the long heating time involved with the growth occurring along the c-axis, Fig. 9(b). Similar structures were reported by Begin et al. when they synthesized HfB2 through mechanical activation.
Figure 9. (a) Field-emission gun scanning electron microscopy of HBeC heat treated at 1300°C for 25 h showing rod shaped particles, (b) Focused dual ion beam and electron backscattered diffraction image of “a.”
Download figure to PowerPoint
(2) Effect of Varying the Hf:B and Hf:C Molar Ratio
No HfB2 formation was observed after heat treating HBe to 1600°C for 5 h; the end product consisted of HfO2 and B2O3, Fig. 10(a); as identified by EDX. This confirms that, without carbon, no reduction reaction occurs in the system, i.e., carbon plays an important role in creating the reducing conditions required for HfB2 formation. In the presence of excess carbon and slight excess boron, HBeCe, the end product contained both HfB2 and ~0.19%, free carbon, as determined by elemental CHN analysis, Fig. 10(b). Interestingly, the size of the HfB2 particles was only 20–30 nm. On the other hand, if both excess boron and carbon were used, HBe1Ce, then boron carbide was formed; Fig. 10(c) and the HfB2 particles were 1–2 μm in size. The presence of the free carbon in the HBeCe system thus appears to act as a physical barrier, preventing particle growth and yielding much finer HfB2 compared with the other powders synthesized, though attempts to remove the residual carbon without chemically degrading the nano HfB2 failed. Without free carbon, the particles are free to grow in size during the heat treatment. Similar results were reported by Krishnarao et. al. during TiB2 synthesis. HfC formation occurred only when the boron source, i.e., Hf:B < 3.8, was present in deficient amounts, see Fig. 5. The XRD of HB and HBeCe powders, Fig. 11, do not give any valuable information regarding the presence of amorphous B2O3 and amorphous carbon. Reactions (2) and (3) represent the excess C and B scenarios, respectively:With slight excess boron excess carbon
With excess boron and excess carbon
(3) Reaction Mechanism
For any reaction to be thermodynamically favorable and spontaneous the Gibbs free energy (ΔG) of the reaction should be negative. For reaction (1a),
ΔG is negative only at temperature >1523°C, as calculated using Factsage 6.1, however, it has been shown that HfB2 forms at a temperature as low as 1300°C. To explain this difference the mechanism by which HfO2, B2O3, and C react to form HfB2 was studied in detail.
Hence, based on thermodynamic calculations, the only reaction that has a realistic chance of leading to HfB2 formation at ~1300°C is reaction (6). Note that at higher temperatures, e.g., above ~1450°C, there may be more than one reaction path. The difference in the experimental and theoretical temperatures was due to the fact that the calculations assumed steady-state conditions and standard state values, whereas the actual reactions took place in a dynamic state.
If the mechanism of formation of HfB2 were to proceed through reaction (4), then HfC should form at a much lower temperature than HfB2. From Fig. 4, it can be seen that HfC peaks did not appear even by 1500°C, though in theory this could be due to the HfC being consumed immediately as it is formed. To understand this better precursor powder HC, made from Hf and C sources, was prepared and heated from 600°C to 1500°C with 0.1 h dwell and subjected to XRD analysis. As seen from Fig. 12, HfO2 formation occurred at 650°C, Fig. 4, but HfC peaks only began to appear at 1500°C. Blum et al. observed that there was a kinetic preference and a lower threshold temperature for the onset of the reaction of Hf with B powder compared with the reaction with C powder. The present work shows that the same also occurs for the reaction between HfO2 and B or C. Hence, it is concluded that HfC is not an intermediate product during HfB2 synthesis, suggesting that reaction (4) is not the correct formation mechanism.
When the HBe1Ce precursor powder was heat treated there was clear evidence of HfB2 formation at 1300°C, as for the HBeC. However, although no boron carbide phases were detected by XRD, even after heat treatment at 1600°C for 2 h, Fig. 13, it was detected by electron microscopy, Fig. 10(c), and EDX analysis revealed the presence of both boron and carbon in the faceted phase. The amount of the boron carbide phase was therefore below the level for XRD detection, especially given the low X-ray density of boron carbide in comparison with hafnium-based compounds.
Figure 14 shows the XRD pattern of the Be1Ce precursor powder heat treated at different temperatures in a scan range 2θ = 30°–40°. It is evident that a boron carbide peak does not appear until 1450°C and hence it is very unlikely to be an intermediate product in HfB2 formation, unless there are some very significant, and unknown, catalytic effects occurring when Hf-based compounds are present.
This may possibly suggest that, at the end of reactions (7)–(9), the products react with HfO2 to form HfB2 before reaction (10) can occur. This is the basis of reaction (6), according to which HfB2 is formed by the reaction between HfO2 and boron suboxides and/or amorphous boron. Due to the limitations of experimental conditions, it was difficult to detect B2O2, B2O, or amorphous B in the intermediate reaction products. Hence, to shed further light on the issue, the experiments were performed involving amorphous boron powder as the source of boron.
High-temperature DTA curves for the precursors HBeC and HBaC, showed similar trends, Fig. 15. The total weight loss for HBaC was 34 wt% whilst that for HBeC was 51 wt%, as mentioned earlier. The difference in the total weight loss may be due to the loss of B2O3 in the HBeC system. The weight loss for the HBaC system occurred up to 500°C due to the loss of physisorbed and chemisorbed water and degradation of the phenolic resin, whilst that for HBeC occurred up to 700°C due to the presence of B2O3 and excess resin. The weight losses remained stable up to 1275°C in both cases and then rapidly decreased and stabilized toward 1600°C.
The endothermic peaks depicting the boro/carbothermal reduction reactions at 1275°C and 1470°C are identical for both precursor powders, indicating that the reaction paths were similar for the two. This could only mean that either the B2O3 had dissociated into B during the course of the HfB2 formation in the HBeC precursor powder or the amorphous B in the HBaC precursor had oxidized into B2O3. As the reaction was carried out in a reducing atmosphere in the presence of carbon, the likelihood for the latter is extremely small.
This result, combined with the Gibbs free energy value, indicates that the formation of HfB2 at 1300°C is most likely to occur through the formation of amorphous boron or boron suboxides as the intermediate product [reaction (6)], though, as indicated earlier, at higher temperatures there could be more than one reaction mechanism taking place simultaneously.