One-Pot Synthesis of Alumina-Titanium Diboride Composite Powder at Low Temperature

Alumina-titanium diboride (Al2O3-TiB2) composite powders were synthesised via aluminothermic reduction of TiO2 and B2O3, mediated by a molten chloride salt (NaCl, KCl, or MgCl2). The effects of salt type, initial batch composition, and firing temperature/time on the phase formation and overall reaction extent were examined. Based on the results and equilibrium thermodynamic calculations, the mechanisms underpinning the reaction/synthesis processes were clarified. Given their evaporation losses at test temperatures, appropriately excessive amounts of Al and B2O3 are needed to complete the synthesis reaction. Following this, phase-pure Al2O3-TiB2 composite powders composed of 0.3–0.6 μm Al2O3 and 30–60 nm TiB2 particles were successfully fabricated in NaCl after 5 h at 1050 °C. By increasing the firing temperature to 1150 °C, the time required to complete the synthesis reaction could be reduced to 4 h, although the sizes of Al2O3 and TiB2 particles in the resultant phase pure composite powder increased slightly to 1–2 μm and 100–200 nm, respectively.


Introduction
Alumina (Al 2 O 3 ) is a representative high performance ceramic material possessing numerous superior properties. Apart from its high melting point, it exhibits high hardness and mechanical strength, excellent wear and chemical resistances, as well as good electrical/thermal insulation [1]. These, along with its ready availability, make Al 2 O 3 an attractive ceramic material applicable to a wide range of structural and functional applications, in, e.g., wear-resistant components, high-speed cutting tools, armors, refractory linings/crucibles, electrical and chemical insulators, and artificial hip joints [2][3][4]. Unfortunately, it suffers from low fracture toughness and poor thermal shock resistance [5,6], negatively affecting its service performance and life. One of the commonly adopted strategies for overcoming this is to make a composite [7], by combining it with another reinforcement phase, e.g., particles/platelets, or fibrous phases such as carbon fibres and carbon nanotubes. Titanium diboride (TiB 2 ) stood out as one of the reinforcement phases, owing to its good structural and thermodynamic compatibility with Al 2 O 3 [8,9] and its excellent properties such as high melting point, high hardness/elastic modulus, relatively low density, and good thermal/electrical conductivities [10]. Previous studies found that incorporation of TiB 2 particles into Al 2 O 3 conferred much improved hardness, strength, fracture toughness and electrical conductivity on the resultant Al 2 O 3 -TiB 2 composite materials [11][12][13][14][15][16], making them suitable for a variety of demanding applications, e.g., in electrodes, cutting tools, wear parts, lightweight armors, high-temperature/glow-plug heaters, and heat exchangers [14,17,18]. To fabricate high performance Al 2 O 3 -TiB 2 composites, it is key to prepare and use high quality counterpart powders. Conventionally, such composite powders were prepared by directly mixing pre-synthesised TiB 2 and Al 2 O 3 powders in a ball mill [11,12,[14][15][16][17]19,20]. Unfortunately, the final bulk composites exhibited unsatisfactory service performance arising mainly from poor dispersion and distribution of TiB 2 in Al 2 O 3 . To overcome this, in-situ formation of the two constituent phases was suggested [21]. For example, a selfpropagating high-temperature synthesis (SHS) method was attempted to form in-situ Al 2 O 3 and TiB 2 from powder mixtures of Al, TiO 2 /Ti, and B 2 O 3 /B/H 3 BO 3 [18,[22][23][24][25][26][27][28][29][30][31][32]. However, this method suffered from several weaknesses; in particular, difficulties to control the synthesis process and synthesise phase pure materials. Apart from SHS, mechanochemical synthesis [8,9,[33][34][35], and milling assisted sol-gel process [36] were investigated. With these two techniques, a long processing time was required, and the resultant powders were often heavily agglomerated, and contaminated during prolonged milling.
In recent years, our group has successfully synthesised several ultra-high temperature boride powders via magnesiothermic reduction of oxide precursors in a molten salt [37][38][39]. In this work, such a molten salt synthesis (MSS) technique was further developed to synthesise high-quality Al 2 O 3 -TiB 2 composite powders at relatively low temperatures. The effects of processing factors such as salt type, initial batch composition, and firing temperature/time on the synthesis process were examined, and the synthesis conditions optimised. Based on the results, and thermodynamic calculations, the mechanisms underpinning the reaction/synthesis process were proposed.

Materials and Methods
The TiO 2 (≥99%, 100-300 nm), B 2 O 3 (99.98%) and Al (99.5%,~44 µm) powders were used as the main starting materials, along with KCl (>99%), NaCl (>99%) and anhydrous MgCl 2 (>98%). All of these materials except for Al (Alfa Aesar, Lancashire, UK) were from Sigma-Aldrich (Gillingham, UK). The three main starting materials were mixed in stoichiometric (indicated by the overall Reaction (1)) or nonstoichiometric proportions (with excess Al and B 2 O 3 ), in a mortar and pestle, and further combined with 5 times (by weight) of each of the salts. The final powder mix was contained in an alumina crucible and heated in an argon protected tube furnace to a temperature between 850 and 1150 • C and held for 4-6 h. After cooling to room temperature, the reacted powder coexisting with the residual salt was subjected to repeated hot water washing. The remaining salt-free product powder was dried overnight at 80 • C before being subjected to detailed characterisations.
Phase formation and reaction extent in samples were evaluated based on their Xray diffraction (XRD) spectra recorded by an X-ray diffractometer (D8 Advance, Bruker, Germany) operated at 40 mA/40 kV (using CuKα radiation), and at a scan rate of 2 • (2θ) min −1 , and the ICDD cards used are: TiB 2 (35-  . Samples were also gold-or carbon-coated for further microstructural examinations using a scanning electron microscope (SEM, Nova 600, FEI, Hillsboro, OR, USA) and a transmission electron microscope (TEM, JEM 2100, JEOL, Tokyo, Japan), linked with energy-dispersive X-ray spectroscopy (EDS, Oxford Instrument, Oxford, UK).
To assist in clarification of the mechanisms underpinning the whole reaction/synthesis process, thermodynamic calculations were carried out using the commercial FactSage package [40]. The Gibbs energy values for each of the intermediate reactions and the overall reaction (Reaction (1)), as a function of temperature, were calculated.   -TiB 2 formation, and the better effect of NaCl than KCl on accelerating the overall reaction. As discussed previously [37][38][39], several processing parameters-in particular, the solubility and mobility of reaction species in the molten salt-play crucial roles in MSS. Unfortunately, these key processing parameters are not available, so it is difficult to figure out the reasons behind the better accelerating effect of NaCl in the present case. When NaCl was replaced with MgCl 2 , the peak heights of Al 2 O 3 and TiB 2 further increased (Figure 1c), and no Al 18 B 4 O 33 , Al 3 Ti or Ti 2 O 3 was detected, suggesting much enhanced extents of Reaction (1). Nevertheless, an undesirable by-product phase, magnesium aluminate spinel (MgAl 2 O 4 ), was formed in this case, due to likely the reaction between the formed Al 2 O 3 and the MgCl 2 (Reaction (2) [41]. In the three cases, except for α-Al 2 O 3 , no other transition aluminas were detected at such a low firing temperature. The reason for this was not clear, but it could be due to the locally enhanced salt bath temperature arising from the exothermic reactions, as suggested by the thermodynamic calculations presented later (see Section 4 below). Given the better accelerating effect of NaCl than KCl, and the formation of undesired by-product MgAl 2 O 4 in the case of using MgCl 2 , NaCl was chosen specifically to form a liquid reaction medium for the following MSS investigations. 3 (2)

Results and Preliminary Discussion
Materials 2021, 14, 3 of 11 Figure 1 shows XRD patterns of stoichiometric samples after 4 h firing in different salts at 850 °C. After firing in KCl, α-Al2O3 and TiB2 were identified as the primary phases in the sample (Figure 1a). Some intermediate Al18B4O33 and Al3Ti were also detected, along with a trace of Ti2O3, indicating the low extents of Reaction (1) and Al2O3-TiB2 formation. Replacing KCl with NaCl ( Figure 1b) led to an evident increase in the relative peak heights of both Al2O3 and TiB2, and a decrease in those of Al3Ti, while those of intermediate Al18B4O33 and Ti2O3 changed little, demonstrating the enhanced extents of Al2O3-TiB2 formation, and the better effect of NaCl than KCl on accelerating the overall reaction. As discussed previously [37][38][39], several processing parameters-in particular, the solubility and mobility of reaction species in the molten salt-play crucial roles in MSS. Unfortunately, these key processing parameters are not available, so it is difficult to figure out the reasons behind the better accelerating effect of NaCl in the present case. When NaCl was replaced with MgCl2, the peak heights of Al2O3 and TiB2 further increased (Figure 1c), and no Al18B4O33, Al3Ti or Ti2O3 was detected, suggesting much enhanced extents of Reaction (1). Nevertheless, an undesirable by-product phase, magnesium aluminate spinel (MgAl2O4), was formed in this case, due to likely the reaction between the formed Al2O3 and the MgCl2 (Reaction (2) [41]. In the three cases, except for α-Al2O3, no other transition aluminas were detected at such a low firing temperature. The reason for this was not clear, but it could be due to the locally enhanced salt bath temperature arising from the exothermic reactions, as suggested by the thermodynamic calculations presented later (see Section 4 below). Given the better accelerating effect of NaCl than KCl, and the formation of undesired by-product MgAl2O4 in the case of using MgCl2, NaCl was chosen specifically to form a liquid reaction medium for the following MSS investigations.

Influence of Salt Type on Al2O3-TiB2 Formation
4Al2O3 + 3MgCl2 = 3MgAl2O4 + 2AlCl3 (2)      (Figure 2d), indicating the adverse effect of such further temperature increase on the Al 2 O 3 -TiB 2 formation. This, according to our previous studies on MSS of transition metal borides [37][38][39], was due to probably evaporation loss of Al at relatively high reaction temperatures [42]. To confirm this, the influence of using excess Al on the Al 2 O 3 -TiB 2 formation was further investigated, as described and discussed next.  (Figure 2c), the relative peak heights of Al2O3 and TiB2 increased significantly. Meanwhile, Al3Ti disappeared, and Ti2O3 changed little, but the peak heights of Al18B4O33 increased. These results implied that increasing the firing temperature favoured not only the Al2O3-TiB2 formation but also the Al18B4O33 formation. Further increasing the temperature to 1150 °C led to a further increase in the peak heights of Al18B4O33 and Ti2O3 but decrease in those of Al2O3 and TiB2 (Figure 2d), indicating the adverse effect of such further temperature increase on the Al2O3-TiB2 formation. This, according to our previous studies on MSS of transition metal borides [37][38][39], was due to probably evaporation loss of Al at relatively high reaction temperatures [42]. To confirm this, the influence of using excess Al on the Al2O3-TiB2 formation was further investigated, as described and discussed next.  Figure 3 illustrates the influence of excess Al on phase evolution in samples after 4 h firing in NaCl at 1150 °C. The use of 20 wt% more Al resulted in considerable increase in the peak heights of Al2O3 and TiB2 and decrease in those of Al18B4O33 and Ti2O3 ( Figure  3a,b). By further increasing excess Al to 25 wt%, the peak heights of Al2O3 and TiB2 were further increased. Meanwhile, Ti2O3 was completely eliminated, and only minor Al18B4O33 remained (Figure 3c), illustrating positive effects from the Al compensation. Nevertheless, the formation of phase-pure Al2O3-TiB2 was still not achieved.

Influence of Excess Al on Al2O3-TiB2 Formation
To eliminate Al18B4O33, the excess Al was further increased to 30 wt%. Unfortunately, some Al3Ti was formed instead (Figure 3d), implying that there was no sufficient supply of B in the system, due to probably evaporation loss of B2O3 at such a relatively high temperature [43,44].   (Figure 3a,b). By further increasing excess Al to 25 wt%, the peak heights of Al 2 O 3 and TiB 2 were further increased. Meanwhile, Ti 2 O 3 was completely eliminated, and only minor Al 18 B 4 O 33 remained (Figure 3c), illustrating positive effects from the Al compensation. Nevertheless, the formation of phase-pure Al 2 O 3 -TiB 2 was still not achieved.
To eliminate Al 18 B 4 O 33 , the excess Al was further increased to 30 wt%. Unfortunately, some Al 3 Ti was formed instead (Figure 3d), implying that there was no sufficient supply of B in the system, due to probably evaporation loss of B 2 O 3 at such a relatively high temperature [43,44]. Materials 2021, 14, 5 of 11

Influence of Excess B2O3 on Al2O3-TiB2 Formation
To confirm and address the issue with B2O3 evaporation loss, the influence of using excess B2O3 (along with 30 wt% excess Al) on the Al2O3-B2O3 formation in the sample fired at 1150 °C for 4 h was further investigated as an example. When 10 wt% excess B2O3 was used, the peak heights of both Al2O3 and TiB2 increased, whereas those of Al3Ti decreased considerably (Figure 4b), compared with in the case of stoichiometric sample (Figure 4a). Further increasing the excess B2O3 to 20 wt% led to complete elimination of Al3Ti, and the formation of phase pure Al2O3-TiB2 (Figure 4c).

Influence of Excess B 2 O 3 on Al 2 O 3 -TiB 2 Formation
To confirm and address the issue with B 2 O 3 evaporation loss, the influence of using excess

Influence of Excess B2O3 on Al2O3-TiB2 Formation
To confirm and address the issue with B2O3 evaporation loss, the influence of using excess B2O3 (along with 30 wt% excess Al) on the Al2O3-B2O3 formation in the sample fired at 1150 °C for 4 h was further investigated as an example. When 10 wt% excess B2O3 was used, the peak heights of both Al2O3 and TiB2 increased, whereas those of Al3Ti decreased considerably (Figure 4b), compared with in the case of stoichiometric sample (Figure 4a). Further increasing the excess B2O3 to 20 wt% led to complete elimination of Al3Ti, and the formation of phase pure Al2O3-TiB2 (Figure 4c).

Influence of Firing Time on Al 2 O 3 -TiB 2 Formation and Further Optimisation of Synthesis Condition
To reveal the influence of firing time on the Al 2 O 3 -TiB 2 formation and further optimise the synthesis condition, samples containing 30 wt% excess Al and 20 wt% excess B 2 O 3 were fired in NaCl at 1050 • C for different time periods, and then analysed by XRD ( Figure 5). As shown in Figure 5a,b, with increasing the time at 1050 • C from 4 to 5 h, Al 18 B 4 O 33 disappeared, whereas the peak heights of Al 3 Ti increased. Further extending the time to 6 h resulted in little change in the XRD pattern (Figure 5c), and Al 3 Ti was still detected. This, as mentioned above (Section 3.4), indicated that 20 wt% excess B 2 O 3 might not be sufficient to compensate for the evaporation loss of B 2 O 3 under this condition, in other words, more excess B 2 O 3 had to be used. As verified by Figure 6, increasing excess B 2 O 3 from 20 wt% to 25 wt% led to significant decrease in the peak heights of Al 3 Ti, and further increasing the excess amount to 30 wt%, all the intermediate phases including Al 3 Ti disappeared, and phase pure Al 2 O 3 -TiB 2 powder was finally obtained.

Influence of Firing Time on Al2O3-TiB2 Formation and Further Optimisation of Synthesis Condition
To reveal the influence of firing time on the Al2O3-TiB2 formation and further optimise the synthesis condition, samples containing 30 wt% excess Al and 20 wt% excess B2O3 were fired in NaCl at 1050 °C for different time periods, and then analysed by XRD ( Figure 5). As shown in Figure 5a,b, with increasing the time at 1050 °C from 4 to 5 h, Al18B4O33 disappeared, whereas the peak heights of Al3Ti increased. Further extending the time to 6 h resulted in little change in the XRD pattern (Figure 5c), and Al3Ti was still detected. This, as mentioned above (Section 3.4), indicated that 20 wt% excess B2O3 might not be sufficient to compensate for the evaporation loss of B2O3 under this condition, in other words, more excess B2O3 had to be used. As verified by Figure 6, increasing excess B2O3 from 20 wt% to 25 wt% led to significant decrease in the peak heights of Al3Ti, and further increasing the excess amount to 30 wt%, all the intermediate phases including Al3Ti disappeared, and phase pure Al2O3-TiB2 powder was finally obtained.

Influence of Firing Time on Al2O3-TiB2 Formation and Further Optimisation of Synthesis Condition
To reveal the influence of firing time on the Al2O3-TiB2 formation and further optimise the synthesis condition, samples containing 30 wt% excess Al and 20 wt% excess B2O3 were fired in NaCl at 1050 °C for different time periods, and then analysed by XRD ( Figure 5). As shown in Figure 5a,b, with increasing the time at 1050 °C from 4 to 5 h, Al18B4O33 disappeared, whereas the peak heights of Al3Ti increased. Further extending the time to 6 h resulted in little change in the XRD pattern (Figure 5c), and Al3Ti was still detected. This, as mentioned above (Section 3.4), indicated that 20 wt% excess B2O3 might not be sufficient to compensate for the evaporation loss of B2O3 under this condition, in other words, more excess B2O3 had to be used. As verified by Figure 6, increasing excess B2O3 from 20 wt% to 25 wt% led to significant decrease in the peak heights of Al3Ti, and further increasing the excess amount to 30 wt%, all the intermediate phases including Al3Ti disappeared, and phase pure Al2O3-TiB2 powder was finally obtained.

Microstructure of Al 2 O 3 -TiB 2 Product Powder
As demonstrated in Figures 4c and 6c and discussed above, phase pure Al 2 O 3 -TiB 2 powder could be formed in NaCl after 5 h firing at 1050 • C (using 30 wt% excess Al and 30 wt% excess B 2 O 3 ), or 4 h firing at 1150 • C (using 30 wt% excess Al and 20 wt% excess B 2 O 3 ), which was further confirmed by microstructural characterisation. Figure 7, as an example, gives TEM images of the product powder prepared under the former condition, along with the corresponding EDS, showing the coexistence of submicron-sized Al 2 O 3 particles (0.3-0.6 µm) and nanosized TiB 2 particles (30-60 nm). Except these two phases, no other impurity phases were seen, as already revealed by XRD (Figure 6c). Figure 8 further displays SEM images of the product powder formed under the latter condition mentioned above. The formation of phase pure Al 2 O 3 -TiB 2 in this case was also confirmed by EDS, along with XRD ( Figure 4c). However, due to the higher synthesis temperature, Al 2 O 3 and TiB 2 particles in the composite powder became slightly bigger (1-2 µm Al 2 O 3 and 100-200 nm TiB 2 ). 14, 7 of 11

Microstructure of Al2O3-TiB2 Product Powder
As demonstrated in Figures 4c and 6c and discussed above, phase pure Al2O3-TiB2 powder could be formed in NaCl after 5 h firing at 1050 °C (using 30 wt% excess Al and 30 wt% excess B2O3), or 4 h firing at 1150 °C (using 30 wt% excess Al and 20 wt% excess B2O3), which was further confirmed by microstructural characterisation. Figure 7, as an example, gives TEM images of the product powder prepared under the former condition, along with the corresponding EDS, showing the coexistence of submicron-sized Al2O3 particles (0.3-0.6 μm) and nanosized TiB2 particles (30-60 nm). Except these two phases, no other impurity phases were seen, as already revealed by XRD (Figure 6c). Figure 8 further displays SEM images of the product powder formed under the latter condition mentioned above. The formation of phase pure Al2O3-TiB2 in this case was also confirmed by EDS, along with XRD ( Figure 4c). However, due to the higher synthesis temperature, Al2O3 and TiB2 particles in the composite powder became slightly bigger (1-2 μm Al2O3 and 100-200 nm TiB2).

Further Discussion and Reaction/Synthesis Mechanism
Based on the results presented in Figures 1-8 and the preliminary discussion above, as well as on the thermodynamic calculations (Figure 9), the reaction/synthesis mechanisms (taking the case of using NaCl as an example) could be proposed; they are discussed as follows.
The firing temperatures (850-1150 °C) were above the melting points of NaCl, B2O3 and Al. So, at these temperatures, they all melted, forming a liquid NaCl pool, and B2O3 and Al liquid droplets. Then, the latter two would start to react with each other in the molten NaCl medium, forming B and Al2O3, according to Reaction (3).

Further Discussion and Reaction/Synthesis Mechanism
Based on the results presented in Figures 1-8 and the preliminary discussion above, as well as on the thermodynamic calculations (Figure 9), the reaction/synthesis mechanisms (taking the case of using NaCl as an example) could be proposed; they are discussed as follows.
. Figure 8. SEM images (a,b) and corresponding EDS (c,d) of the Al2O3-TiB2 composite powder resultant from 4 h firing in NaCl at 1150 °C.

Further Discussion and Reaction/Synthesis Mechanism
Based on the results presented in Figures 1-8 and the preliminary discussion above, as well as on the thermodynamic calculations (Figure 9), the reaction/synthesis mechanisms (taking the case of using NaCl as an example) could be proposed; they are discussed as follows.
The firing temperatures (850-1150 °C) were above the melting points of NaCl, B2O3 and Al. So, at these temperatures, they all melted, forming a liquid NaCl pool, and B2O3 and Al liquid droplets. Then, the latter two would start to react with each other in the molten NaCl medium, forming B and Al2O3, according to Reaction (3).  The firing temperatures (850-1150 • C) were above the melting points of NaCl, B 2 O 3 and Al. So, at these temperatures, they all melted, forming a liquid NaCl pool, and B 2 O 3 and Al liquid droplets. Then, the latter two would start to react with each other in the molten NaCl medium, forming B and Al 2 O 3 , according to Reaction (3).
As discussed, and confirmed in our previous studies [37][38][39], the formed B and Ti (Reactions (3), (4) and (6)) were slightly dissoluble in the molten chloride salts. So, the dissolved B and Ti would diffuse through the molten salt medium and reacted with each other, forming the other desirable phase of TiB 2 (Reaction (8) Ti + 3Al = Al 3 Ti (7) Ti + 2B = TiB 2 (8) B dissolved in the molten salt would also diffuse through the salt medium and react respectively with the Al 3 Ti produced from Reaction (7), and the unreacted raw materialTiO 2 , forming additional TiB 2 (Reactions (9) and (10)). Reaction (9) was mainly responsible for the complete elimination of this intermediate phase under the optimal conditions (Figures 4 and 6).

Conclusions
Low temperature synthesis of Al 2 O 3 -TiB 2 composite powders via aluminothermic reduction of TiO 2 and B 2 O 3 in a molten chloride salt (KCl, NaCl or MgCl 2 ) was investigated. Processing parameters including salt type, initial batch composition, and firing temperature/time played important roles in the synthesis process. Phase pure Al 2 O 3 -TiB 2 nanocomposite powder could be prepared in NaCl after 5 h firing at 1050 • C, when 30 wt% excess Al and 30 wt% excess B 2 O 3 were used. The particle sizes of Al 2 O 3 and TiB 2 coexisting in the as-prepared composite powder were estimated to be 0.3-0.6 µm and 30-60 nm, respectively. By increasing the temperature to 1150 • C and using 30 wt% excess Al and 20 wt% excess B 2 O 3 , phase pure Al 2 O 3 -TiB 2 composite powder also could be prepared in NaCl after only 4 h, although sizes of Al 2 O 3 and TiB 2 particles coexisting in the resultant composite power increased slightly to 1-2 µm and 100-200 nm, respectively.
Author Contributions: Experiment and initial draft, K.B. and X.L.; conception, experiment design and thermodynamic calculation, K.B., X.L., J.C. and Q.J.; review and editing, X.L., Q.J. and S.Z.; supervision, S.Z. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the National Natural Science Foundation of China (Grant No. 51902180), and the EPSRC Centre for Doctoral Training in Electromagnetic Metamaterials (EP/L015331/1) at the University of Exeter.