Combustion Synthesis of NbB2–Spinel MgAl2O4 Composites from MgO-Added Thermite-Based Reactants with Excess Boron

The formation of NbB2–MgAl2O4 composites from the MgO-added thermite-based reaction systems was investigated by self-propagating high-temperature synthesis (SHS). Two thermite mixtures, Nb2O5/B2O3/Al and Nb2O5/Al, were, respectively, adopted in Reactions (1) and (2). The XRD analysis confirmed the combination of Al2O3 with MgO to form MgAl2O4 during the SHS process and that excess boron of 30 atom.% was required to yield NbB2–MgAl2O4 composites with negligible NbB and Nb3B4. The microstructure of the composite reveals that rod-shaped MgAl2O4 crystals are closely interlocked and granular NbB2 are embedded in or scattered over MgAl2O4. With the addition of MgAl2O4, the fracture toughness (KIC) of 4.37–4.82 MPa m1/2 was obtained for the composites. The activation energies Ea = 219.5 ± 16 and 167.9 ± 13 kJ/mol for Reactions (1) and (2) were determined from combustion wave kinetics.


Introduction
Transition metal (IVB and VB) diborides, such as TiB 2 , ZrB 2 , NbB 2 , and TaB 2 , have been referred to as ultra-high-temperature ceramics (UHTCs). Transition metal diborides crystallize in the hexagonal A1B 2 type structure (space group P6/mmm) with c/a ratio close to unity. In this arrangement, the hexagonal nets of metal atoms and triangle nets of pure boron atoms are alternately stacked along the c axis [1]. Besides their melting points exceeding 3000 • C, they possess a unique combination of high hardness, thermal conductivity, electrical conductivity, excellent chemical stability, corrosion resistance, and thermal shock resistance [2][3][4][5]. They have found a broad range of applications in the mechanical, automobile, aerospace industries, etc. [5]. To improve the refractory properties of metal borides and carbides, Al 2 O 3 or MgAl 2 O 4 (magnesium aluminate spinel) has been considered as an additive. Extensive studies have been conducted on understanding the production and characteristics of different metal borides reinforced by Al 2 O 3 [6][7][8][9][10], but relatively few studies have focused on the boride-MgAl 2 O 4 composite [11].
MgAl 2 O 4 is of particular interest due to its exceptional mechanical, thermal, and optical properties, such as high melting point (2135 • C), high hardness (16 GPa), relatively low density (3.58 g/cm 3 ), high mechanical strength (135-216 MPa), good transmittance in the wavelength range of 0.25 to 5.0 µm, high thermal shock resistance, high chemical inertness, low dielectric constant, and low thermal expansion coefficient [12,13]. MgAl 2 O 4 spinel has a close-packed face-centered cubic (fcc) structure of space group Fd3m (number 227). There are eight MgAl 2 O 4 units per cubic cell. Mg and Al cations occupy 1/8 of the tetrahedral sites and 1/2 of the octahedral sites and there are 32 oxygen ions in the unit cell. In the normal spinel, all Al 3+ ions are in octahedral coordination with local symmetry D 3d
where x and y are stoichiometric coefficients signifying the mole number of NbB 2 formed per unit mole of MgAl 2 O 4 in Reactions (1) and (2), respectively. The experiments of Reaction (1) were performed with x = 0.6-1.0. The increase of x raises the amount of Nb 2 O 5 but reduces that of B 2 O 3 in the reactant mixture. Under this condition, more amorphous boron is added to make up for the decrease of boron provided from B 2 O 3 . Samples of Reaction (2) were formulated with y = 1.2-1.8. The increase of y augments elemental Nb and B for the production of a larger amount of NbB 2 , but has no change in the molar quantity of thermite reagents, Nb 2 O 5 and Al, in Reaction (2). A combination of Reactions (1) and (2) renders this study feasible to obtain products with a molar proportion of NbB 2 /MgAl 2 O 4 from 0.6 to 1.8. In addition, the SHS reactions with test specimens containing excess boron of 20 and 30 atom.% were conducted to examine the extent of boron loss during combustion and to compensate for the relatively low purity (92%) of amorphous boron used in this study. It should be noted that boron has high hardness, great stability in extreme environments, and good resistance to heat. It has several forms, the most common of which is amorphous boron, and is unreactive to oxygen, water, acids, and alkalis. While barely reactive at room temperature, boron reacts strongly at high temperature with metals to form borides. Moreover, its reducing properties allow it to react with numerous compounds and, as in the case of oxygenated or halogenated compounds, in a violent manner [26].
In Reaction (1), the aluminothermic reduction of Nb 2 O 5 (the reaction enthalpy, ∆H r = −536 kJ/mol of Al 2 O 3 and the adiabatic temperature, T ad = 2756 K) is more energetic than that of B 2 O 3 (∆H r = −403.8 kJ/mol of Al 2 O 3 and T ad = 2315 K) [27,28]. The intermetallic reaction of Nb + 2B to yield NbB 2 with formation enthalpy ∆H f = −175.3 kJ/mol and T ad = 2315 K is exothermically comparable to the B 2 O 3 + 2Al reaction [29]. The combination reaction between MgO and Al 2 O 3 to form MgAl 2 O 4 is weakly exothermic with ∆H r = −35.6 kJ [23]. This means that with the increase of NbB 2 content (the x value), two opposing effects govern the combustion exothermicity of Reaction (1). As far as Reaction (2) is concerned, the increase of y for the formation of a larger NbB 2 content has a dilution effect on combustion. To elucidate the combustion exothermicity, calculation of T ad of Reactions (1) and (2) under different stoichiometric coefficients was performed according to the following equation [23,30] with thermochemical data taken from [29].
where ∆H r is the reaction enthalpy at 298 K, n j is the stoichiometric constant, C p and L are the heat capacity and latent heat, and P j refers to the product. Calculations of the adiabatic temperature were based upon the final products of stoichiometric reactions described in R(1) and R (2).
Reactant powders were mixed in a ball mill and then uniaxially compressed in a stainless-steel mold at a pressure of 70-80 MPa to form cylindrical test samples with 7 mm in diameter, 12 mm in height, and a relative density of 60%. The relative density of the test specimen is related to the initial components. The theoretical density (ρ TD ) of the test specimen is calculated from the mass fraction (Y) and density (ρ) of each component through the following equation.
The SHS experiments were conducted in a windowed combustion chamber filled with high-purity (99.99%) argon. The propagation velocity of the combustion wave (V f ) was determined from the time sequence of recorded pictures. The reaction temperature was measured by the Pt/Pt-13%Rh bare wire thermocouple with a bead diameter of 125 µm. A thin ceramic (SiO 2 ) coating is usually to prevent the catalytic effect on the thermocouple in the measurement of gas-phase flame temperature of a combustion mixture involving hydrogen, methane, or propane as the fuel. Because solid-state combustion with argon as the surrounding gas is under investigation, the thermocouple without an inert coating was used by this work. Details of the experimental setup and scheme were described elsewhere [25,31]. After the SHS process, phase constituents of the products were analyzed by an X-ray diffractometer (Bruker D2, Billerica, MA, US) using CuK α radiation. The microstructure of the final product was examined by a scanning electron microscope (Hitachi S3000H, Tokyo, Japan) and elemental proportion was deduced from the energy dispersive spectroscopy (EDS). For Vickers hardness (H v ) and fracture toughness (K IC ) measurement, selected experiments with the reactant compact placing in a steel mold were conducted. Upon the completion of the self-sustaining combustion reaction, densification of the product was carried out by a hydraulic press machine [25]. Figure 1 illustrates a typical series of combustion images obtained by this study, which was recorded from the powder compact of Reaction (2) with y = 1.4. As can be seen in Figure 1, a well-defined combustion wave is established upon ignition and propagates throughout the entire sample in a self-sustaining manner. This demonstrates sufficient reaction exothermicity of the reactant mixture. After combustion, the burned sample essentially retained its original shape.

Combustion Wave Kinetics and Reaction Temperature
Crystals 2020, 10, x FOR PEER REVIEW 4 of 10 well-defined combustion wave is established upon ignition and propagates throughout the entire sample in a self-sustaining manner. This demonstrates sufficient reaction exothermicity of the reactant mixture. After combustion, the burned sample essentially retained its original shape. The influence of the stoichiometric coefficient and excess boron on the flame-front velocity of Reactions (1) and (2) is presented in Figure 2. It was found that for the samples without excess boron, the combustion wave velocity of Reaction (1) increased from 2.5 to 6.6 mm/s with x increasing from 0.6 to 1.0, while that of Reaction (2) decreased from 8.4 to 4.8 mm/s with y in the range from 1.2 to 1.8. To be presented lately, it is believed that the variation of combustion front velocity with the reaction stoichiometry depends mainly on the exothermicity of the SHS process. As also indicated in Figure  2, samples with excess boron of 30 atom.% exhibit lower combustion wave speeds when compared to those without additional boron. This could be attributable to a prolonged sequence of phase evolution of borides in response to the increase of boron. Figure 3 plots several measured sample temperature profiles, which depict a sharp rise signifying the rapid arrival of the combustion wave and a peak value corresponding to the combustion front temperature (Tc). After the progression of the combustion wave, a substantial temperature decline is a consequence of heat loss to the surroundings. As revealed in Figure 3, the peak temperature (Tc = 1516 °C) of Reaction (1) with x = 0.9 is higher than that of x = 0.6 (Tc = 1317 °C) and the combustion front temperature of Reaction (2) decreases from 1606 °C to 1416 °C with an increase of y from 1.2 to 1.8. Based on the experimental measurement, the stoichiometric dependence of the combustion wave temperature is in agreement with that of flame-front velocity.  The influence of the stoichiometric coefficient and excess boron on the flame-front velocity of Reactions (1) and (2) is presented in Figure 2. It was found that for the samples without excess boron, the combustion wave velocity of Reaction (1) increased from 2.5 to 6.6 mm/s with x increasing from 0.6 to 1.0, while that of Reaction (2) decreased from 8.4 to 4.8 mm/s with y in the range from 1.2 to 1.8. To be presented lately, it is believed that the variation of combustion front velocity with the reaction stoichiometry depends mainly on the exothermicity of the SHS process. As also indicated in Figure 2, samples with excess boron of 30 atom.% exhibit lower combustion wave speeds when compared to those without additional boron. This could be attributable to a prolonged sequence of phase evolution of borides in response to the increase of boron.
Crystals 2020, 10, x FOR PEER REVIEW 4 of 10 well-defined combustion wave is established upon ignition and propagates throughout the entire sample in a self-sustaining manner. This demonstrates sufficient reaction exothermicity of the reactant mixture. After combustion, the burned sample essentially retained its original shape. The influence of the stoichiometric coefficient and excess boron on the flame-front velocity of Reactions (1) and (2) is presented in Figure 2. It was found that for the samples without excess boron, the combustion wave velocity of Reaction (1) increased from 2.5 to 6.6 mm/s with x increasing from 0.6 to 1.0, while that of Reaction (2) decreased from 8.4 to 4.8 mm/s with y in the range from 1.2 to 1.8. To be presented lately, it is believed that the variation of combustion front velocity with the reaction stoichiometry depends mainly on the exothermicity of the SHS process. As also indicated in Figure  2, samples with excess boron of 30 atom.% exhibit lower combustion wave speeds when compared to those without additional boron. This could be attributable to a prolonged sequence of phase evolution of borides in response to the increase of boron. Figure 3 plots several measured sample temperature profiles, which depict a sharp rise signifying the rapid arrival of the combustion wave and a peak value corresponding to the combustion front temperature (Tc). After the progression of the combustion wave, a substantial temperature decline is a consequence of heat loss to the surroundings. As revealed in Figure 3, the peak temperature (Tc = 1516 °C) of Reaction (1) with x = 0.9 is higher than that of x = 0.6 (Tc = 1317 °C) and the combustion front temperature of Reaction (2) decreases from 1606 °C to 1416 °C with an increase of y from 1.2 to 1.8. Based on the experimental measurement, the stoichiometric dependence of the combustion wave temperature is in agreement with that of flame-front velocity. A comparison between the calculated adiabatic temperature and measured combustion front temperature of Reactions (1) and (2) is presented in Figure 4, indicative of a consistent variation of Tad and Tc with the reaction stoichiometry. The increase of the combustion temperature of Reaction  Figure 3 plots several measured sample temperature profiles, which depict a sharp rise signifying the rapid arrival of the combustion wave and a peak value corresponding to the combustion front temperature (T c ). After the progression of the combustion wave, a substantial temperature decline is a consequence of heat loss to the surroundings. As revealed in Figure 3, the peak temperature (T c = 1516 • C) of Reaction (1) with x = 0.9 is higher than that of x = 0.6 (T c = 1317 • C) and the combustion front temperature of Reaction (2) decreases from 1606 • C to 1416 • C with an increase of y from 1.2 to 1.8. Based on the experimental measurement, the stoichiometric dependence of the combustion wave temperature is in agreement with that of flame-front velocity.
Crystals 2020, 10, x FOR PEER REVIEW 5 of 10 (1) with increasing NbB2/MgAl2O3 ratio is ascribed to a larger proportion of Nb2O5 to B2O3 in the thermite mixture, since the aluminothermic reduction of Nb2O5 is more exothermic. However, a decline of the combustion temperature of Reaction (2) with NbB2/MgAl2O3 molar ratio confirms the cooling effect on combustion by increasing Nb and B, because the elemental reaction of Nb with B is less energetic than the thermite reaction of Nb2O5 and Al. As shown in Figure 4, the values of Tad are higher by about 350-400 °C than those of Tc. The discrepancy between Tc and Tad might result from considerable heat loss mostly by radiation, substantial boron elimination from the reaction zone, and formation of boride phases different from the stoichiometric composition.   (1) and (2) as a function of the NbB2/MgAl2O4 molar ratio.
The activation energy (Ea) of solid-state combustion was deduced from combustion wave kinetics by constructing a correlation between ln(Vf/Tc) 2 and 1/Tc in a form of linear relationship [32,33]. Figure 5 depicts two sets of experimental data with best-fitted straight lines. From the slopes of straight lines, Ea = 219.5 ± 16 and 167.9 ± 13 kJ/mol were deduced for Reactions (1) and (2), respectively. A larger Ea for Reaction (1) means a higher kinetic barrier in comparison to Reaction (2). This could be caused most likely by the fact that the co-reduction of Nb2O5 and B2O3 by Al is required in Reaction (1) for the synthesis sequence to proceed, but Reaction (2) has only Nb2O5 to be reduced. According to Arrhenius kinetics, the activation energy of the solid-state reaction is governed by the reaction mechanism. In this study, aluminothermic reduction of metal oxides is  (1) and (2) with different stoichiometric coefficients.
A comparison between the calculated adiabatic temperature and measured combustion front temperature of Reactions (1) and (2) is presented in Figure 4, indicative of a consistent variation of T ad and T c with the reaction stoichiometry. The increase of the combustion temperature of Reaction (1) with increasing NbB 2 /MgAl 2 O 3 ratio is ascribed to a larger proportion of Nb 2 O 5 to B 2 O 3 in the thermite mixture, since the aluminothermic reduction of Nb 2 O 5 is more exothermic. However, a decline of the combustion temperature of Reaction (2) with NbB 2 /MgAl 2 O 3 molar ratio confirms the cooling effect on combustion by increasing Nb and B, because the elemental reaction of Nb with B is less energetic than the thermite reaction of Nb 2 O 5 and Al. As shown in Figure 4, the values of T ad are higher by about 350-400 • C than those of T c . The discrepancy between T c and T ad might result from considerable heat loss mostly by radiation, substantial boron elimination from the reaction zone, and formation of boride phases different from the stoichiometric composition. (1) with increasing NbB2/MgAl2O3 ratio is ascribed to a larger proportion of Nb2O5 to B2O3 in the thermite mixture, since the aluminothermic reduction of Nb2O5 is more exothermic. However, a decline of the combustion temperature of Reaction (2) with NbB2/MgAl2O3 molar ratio confirms the cooling effect on combustion by increasing Nb and B, because the elemental reaction of Nb with B is less energetic than the thermite reaction of Nb2O5 and Al. As shown in Figure 4, the values of Tad are higher by about 350-400 °C than those of Tc. The discrepancy between Tc and Tad might result from considerable heat loss mostly by radiation, substantial boron elimination from the reaction zone, and formation of boride phases different from the stoichiometric composition.  The activation energy (Ea) of solid-state combustion was deduced from combustion wave kinetics by constructing a correlation between ln(Vf/Tc) 2 and 1/Tc in a form of linear relationship [32,33]. Figure 5 depicts two sets of experimental data with best-fitted straight lines. From the slopes of straight lines, Ea = 219.5 ± 16 and 167.9 ± 13 kJ/mol were deduced for Reactions (1) and (2), respectively. A larger Ea for Reaction (1) means a higher kinetic barrier in comparison to Reaction (2). The activation energy (E a ) of solid-state combustion was deduced from combustion wave kinetics by constructing a correlation between ln(V f /T c ) 2 and 1/T c in a form of linear relationship [32,33]. Figure 5 depicts two sets of experimental data with best-fitted straight lines. From the slopes of straight lines, E a = 219.5 ± 16 and 167.9 ± 13 kJ/mol were deduced for Reactions (1) and (2), respectively. A larger E a for Reaction (1) means a higher kinetic barrier in comparison to Reaction (2). This could be caused most likely by the fact that the co-reduction of Nb 2 O 5 and B 2 O 3 by Al is required in Reaction (1) for the synthesis sequence to proceed, but Reaction (2) has only Nb 2 O 5 to be reduced. According to Arrhenius kinetics, the activation energy of the solid-state reaction is governed by the reaction mechanism. In this study, aluminothermic reduction of metal oxides is considered as a first step of the SHS process, which is followed by a combination of Al 2 O 3 and MgO to form MgAl 2 O 4 and elemental interactions between Nb and B to produce NbB 2 .
Crystals 2020, 10, x FOR PEER REVIEW 6 of 10 considered as a first step of the SHS process, which is followed by a combination of Al2O3 and MgO to form MgAl2O4 and elemental interactions between Nb and B to produce NbB2.

Phase Composition and Microstructure of As-Synthesized Products
Figure 6a-c displays the XRD patterns of SHS-derived products from Reaction (1) of x = 1.0 without and with excess boron. As shown in Figure 6, MgAl2O4 is identified and niobium borides exist in three phases including NbB2, Nb3B4, and NbB. The formation of MgAl2O4 confirms a combination reaction between pre-added MgO and thermite-produced Al2O3. The presence of NbB and Nb3B4 denotes that the amount of boron is inadequate to transform all the borides into NbB2. For the sample without extra boron, as shown in Figure 6a, NbB is the dominant boride phase. Besides, there is NbO2 detected in the final product, indicative of an incomplete reduction of Nb2O5. For the sample with excess boron of 20 atom.%, Figure 6b indicates that the yield of Nb3B4 and NbB2 is enhanced and NbO2 is no longer detectable. Furthermore, Figure 6c reveals that more NbB2 is formed in the final product obtained from the sample with extra boron of 30 atom.%. Although NbB2 is the dominant boride in Figure 6c, the other two borides, NbB and Nb3B4, are not trivial. This means a substantial boron loss from the samples of Reaction (1) possibly through two different paths. One is the reduction of B2O3 and the other is the borothermal reaction between Nb2O5 and boron. Both reactions could generate gaseous B2O2 and BO, and they might expel from the sample compact.
The effect of excess boron on the formation of borides for Reaction (2) is presented in Figure 7ac. Likewise, excess boron enhanced the production of NbB2. It was found that the improvement is more effective for Reaction (2) than Reaction (1). Figure 7b unveils that NbB2 prevails over NbB and Nb3B4 for the sample with excess boron of 20 atom.%. Moreover, as shown in Figure 7c, NbB and Nb3B4 become negligible in the resulting product from the sample with excess boron of 30 atom.%. This is because Reaction (2) contains no B2O3, only borothermal reduction of Nb2O5 could result in the loss of boron.
Typical microstructures of the fracture surface of the NbB2-MgAl2O4 composites synthesized from Reactions (1) and (2) with 30 atom.% extra boron are illustrated in Figure 8a,b, respectively. As displayed in the micrographs of Figure 8a,b; the long rod-shaped MgAl2O4 crystals form a dense   Figure 6c, the other two borides, NbB and Nb 3 B 4 , are not trivial. This means a substantial boron loss from the samples of Reaction (1) possibly through two different paths. One is the reduction of B 2 O 3 and the other is the borothermal reaction between Nb 2 O 5 and boron. Both reactions could generate gaseous B 2 O 2 and BO, and they might expel from the sample compact.

Phase Composition and Microstructure of As-Synthesized Products
The effect of excess boron on the formation of borides for Reaction (2) is presented in Figure 7a-c. Likewise, excess boron enhanced the production of NbB 2 . It was found that the improvement is more effective for Reaction (2) than Reaction (1). Figure

Conclusions
This study prepared NbB2-MgAl2O4 in situ composites with a molar ratio of NbB2/MgAl2O4 from 0.6 to 1.8 by the SHS process with reducing stages. Within the scope of experimental variables,

Conclusions
This study prepared NbB 2 -MgAl 2 O 4 in situ composites with a molar ratio of NbB 2 /MgAl 2 O 4 from 0.6 to 1.8 by the SHS process with reducing stages. Within the scope of experimental variables, combustion front velocity and temperature increased with NbB 2 content for Reaction (1), because the proportion of Nb 2 O 5 to B 2 O 3 in the thermite mixture increased. On the other hand, Reaction (2) showed a decrease in combustion velocity and temperature as NbB 2 content increased, because of the dilution effect of additional Nb and B on combustion. The activation energies, E a = 219.5 ± 16 and 167.9 ± 13 kJ/mol, were, respectively, deduced for Reactions (1) and (2), suggesting a higher kinetic barrier for Reaction (1). The