In-situ synthesis of TiN and TiB2 compounds during reactive spark plasma sintering of BN–Ti composites

A BN-TiB2-TiN composite was produced via reactive sintering of the hexagonal BN (hBN) with 20 wt% Ti. Spark plasma sintering (SPS) was used as the fabrication method and the sample was characterized by X-ray diffractometry, energy-dispersive X-ray spectroscopy, and scanning electron microscopy. According to the results, the Ti was utterly consumed during the SPS, led to the in-situ TiB2 and TiN0.9 formations. Additionally, the microstructural study revealed the nucleation and growth of new hBN platelets from the initial fine hBN particles. Anyway, the final composite reached a relative density of 95%, because of the remaining free spaces between the hBN platelets. It was found that some nitrogen and boron atoms could leave the TiN and TiB2 microstructures, respectively, and diffuse into the opposing phase. © 2021 The Authors. Published by Synsint Research Group. Boron nitride Titanium Reactive spark plasma sintering In-situ phases Synthesis


Introduction 1.
Hexagonal boron nitride is an advanced ceramic with a crystalline structure similar to graphite, namely a lamellar one [1][2][3][4]. This material owns a unique combination of features, e.g., low elastic modulus, low density, high thermal conductivity, good thermal shock resistivity, high refractoriness, high abrasion resistivity, excellent corrosion resistivity, and significant dielectric and insulation characteristics, as well as good machinability, which makes it a suitable nominee for an extended range of industrial applications, such as thermocouples' insulating sleeves, crucibles, protection tubes, and high-temperature furnaces [5][6][7][8][9][10][11]. Additionally, this material possesses poor overall hardness owing to having the weak inter-layer van der Waals bonding [12]. Nevertheless, having both anisotropic lamellar structure as well as powerful covalent bonding has made the sintering process of this material challenging [13,14]. Although grain coarsening is highly prohibited due to the low diffusion coefficient of hBN, the lamellar structure of this substance results in remaining porosity between the plates [4]. To overcome the mentioned difficulties, the implementation of advanced sintering techniques and also the incorporation of sintering additives can be the solution. The pressureless sintering (PS) [15][16][17][18], hot-pressing (HP) [19][20][21][22][23], and sintering by spark plasma (SPS) [24][25][26][27][28][29] are amongst the most used processes for sintering of ceramic materials. The first route, i.e., pressureless sintering, is considered as a commercial technique, while the utilization of two others can lead to fabricating high-quality compounds thanks to their intrinsic characteristics, namely simultaneous external pressure and comparably low sintering temperature [1,[30][31][32][33]. Moreover, many scientists performed numerous  researches works to improve both qualifications and sinterability of  hBN via introducing an appropriate sintering additive, including Ti, Ni,  TiN, CaB 2 O 4 , MAS, YAG, SiO 2 , MgO, CaO, B 2 O 3 , Al 2 O 3 , Y 2 O 3 , and  so forth [4,12,30,31,[34][35][36][37][38]. Kitiwan et al. [38] studied the densification behavior as well as phase transformation of the SPSed hBN-TiN composites produced at four different sintering temperatures, i.e., 1700, 1800, 1900 and 2000 °C. According to their results, the relative density of samples was linearly increased with rising TiN amount from 10 to 90 vol%. As a result, the highest percentage of relative density (almost 97%) was secured for the sample containing 90% TiN SPSed at 2000 °C. Also, the formation of TiB 2 as an in-situ phase was reported at all sintering temperatures, excluding the lowest one, namely 1700 °C. They also investigated the TiN-TiB 2 -hBN composites fabricated by the SPS route at 1700 °C under 100 MPa for 5 min. In this research, the composites were containing between 0 to 30 vol% BN, while the ratio of TiB 2 /TiN was maintained aound 70:30 (vol%). All samples reached the relative density values in the range of 96-97%; however, the optimum mechanical qualifications were attained at hBN content of 15%. Finally, Olevsky et al. [35] scrutinized the hBN/Ti composites produced via the pressureless sintering route. In terms of the hBN/Ti system with an atomic percent ratio of 2:3, the results endorsed the full consumption of the raw materials and the in-situ synthesis of TiB 2 and TiN compounds (at sintering temperate of 1200 °C for 20 hours). In this research paper, we have tried to assess the impact of incorporating 20 wt% Ti as an additive on the microstructure of hBN. For this aim, the spark plasma sintering method at 1900 °C was engaged to densify the composite specimen. Additionally, energydispersive X-ray spectroscopy, scanning electron microscopy, and Xray diffractometry were employed to investigate the SPSed sample.

Experimental procedure 2.
In this investigation, commercially accessible powders of titanium and boron nitrides were used as the initial materials. A brief of the characteristics of such substances is exhibited in Table 1. First of all, the needed amount of each powder was weighed to fabricate BN-based samples containing 20 wt% Ti as an additive. Subsequently, the powder mixture was ultrasonically dispersed for 40 min in ethanol. In order to remove the already added ethanol as well as applying more efficient mixing, the resulted slurry was heated for two hours at 125 °C using a hot-plate shaker. After that, complete dehumidification was carried out for 24 hours at 100 °C using a universal oven. A uniform powder mixture was attained by smashing the prepared dried slurry and passing it by a screen (100-mesh). Finally, the green powder mixture was poured into a graphite die (diameter of 30 mm) to be densified at 1900 °C under 40 MPa for 10 min in a Nanozint 10i SPS facility (KPF Co., Iran). It is also worth mentioning that the inside of the graphite mold was lined with a soft graphite foil to prohibit any chemical reaction between the die and the mixture. However, such a line was removed via grinding after the SPS process. The relative density of the SPSed composite was measured as a fraction of the bulk density estimated by the Archimedes principle to the theoretical density metered using the rule of mixture. The elemental assessment was performed using DXP-X10P energy-dispersive X-ray spectroscopy (EDS), whereas the phase analysis was carried out using Table 1. The commercial raw materials used in this investigation as well as their relevant information.
Philips, PW1730 X-ray diffractometry (XRD). The former appliance was fitted with a Tescan Mira3 field emission scanning electron microscopy (Czech Republic) used in this study for microstructure assessment both on fracture and polished surfaces. In addition, a software called HSC (Outokumpu) was utilized for thermodynamic evaluations.
Results and discussion 3.

Sintering behavior and thermodynamic evaluation
Based on the electron microscopy images and the XRD spectrums of as-purchased powders (not shown here), it can be noted that the morphology and particle size of each powder is in agreement with the datasheet presented by the manufacturer (Table 1). Additionally, the XRD spectrums imply that both starting materials were highly pure, and as a result, the amounts of possible impurities in these powders were such low that the XRD machine could not detect them. Nevertheless, as is well-known, the presence of oxide compounds on the surface of such powder particles is inevitable thanks to the high reactivity of both titanium and boron with the oxygen element available in the atmosphere. Accordingly, it can be assumed that Ti and BN powders are mainly covered by TiO2 and B 2 O 3 surface oxides, respectively, based on the literature. Oxide impurities may have various roles in a specific sintering system. Briefly, oxide ingredients can either worsen or improve the sinterability of a ceramic-based composite. On the one hand, the presence of surface oxides on the adjacent particles, which are going to coalesce to each other, not only prohibits a powerful bonding between them, but also such contaminations may lead to grain coarsening during the sintering process. The latter phenomenon would be more problematic in the sintering routes with long soaking time, e.g., hot press and pressureless sintering. On the other hand, if the melting temperature of these oxide compounds would be lower than the maximum sintering temperature or they would be able to create a new low melting point ingredient together, a densification mechanism called liquid phase sintering can be activated over the process, boosting the sintering behavior of the composite. In any event, the assessment of the possible chemical interactions between the available phases, as well as recognition of the probable consolidation mechanisms over the sintering route, would be beneficial in finding a clear insight from the sintering behavior of a ceramic-based composite. According to our calculation, the hBN-20 wt% Ti sample reached a relative density of 95.3%, which implies almost 5% remaining porosity in its microstructure. The XRD spectrum of this specimen is presented in Fig. 1. According to this pattern, the Ti content was totally consumed over the SPS process, and two in-situ phases, i.e., TiB2 and TiN 0.9 , were produced. In the following, the chemical reactions between the phases available in the hBN-Ti system will be evaluated. First of all, as noted earlier, the hBN particles are inherently covered by the B 2 O 3 oxide layer. This phase has a low melting point (between 450-510 °C), and as a result, forms a liquid phase (Eq. 1) at the initial steps of the SPS process. This phenomenon, as mentioned before, is called the liquid phase sintering mechanism, which is advantageous in enhancing the sinterability of the sample via particle rearrangement. In short, the existing particles may rearrange in a liquid bed, through which less free space remains amongst them. In addition, the fragmentation of large particles is another densification mechanism at low temperatures obtained as a result of the simultaneous applied pressure. Such a thing is also helpful in more reducing the free space mentioned above. Although B 2 O 3 has a controlling role in the sinterability of the hBN-Ti composite at low temperatures, it evaporates and leaves the system thanks to its low boiling point (1860 °C), as well as the high applied vacuum. This matter, along with the fact that B 2 O 3 cannot chemically interact with other available phases in the system under the current sintering circumstances, reveals that such an oxide does not have any role in the sinterability of hBN-Ti at high temperatures.
Considering the TiO 2 oxide, the issue is utterly different. This oxide not only owns a higher melting point (1843 °C), but it also cannot evaporate and leave the system, due to its high boiling point (2972 °C). In other words, TiO 2 can generate a molten phase at the maximum sintering temperature, participating in filling the remaining porosity. This oxide finally solidifies as an amorphous compound in the microstructure. Similar to B 2 O 3 , TiO 2 cannot react with other existing ingredients and remain unreacted.
By contrast, BN and Ti have a strong reactivity to each other. In line with Eq. 2, Ti and BN ingredients can participate in a chemical reaction, generating in-situ TiB 2 and TiN phases. Such a reaction was evaluated using the HSC chemical package, and it was found that this reaction can easily advance at the current sintering system given its highly negative ∆G° at 1900 °C (almost -376 kJ). Olevsky et al. [35] also reported an equilibrium amongst the TiB 2 , TiN, and BN phases in the ternary system of Ti-B-N. In view of the value calculated for the standard Gibbs free energy of Eq. 1, this fact that all Ti additive was consumed during the sintering process makes sense. However, the reason for identifying the TiN phase as a non-stoichiometric one in the XRD spectrum (Fig. 1) Fig. 2 a-c exhibits the FESEM micrographs from the polished section of the hBN-20 wt% Ti sample at different magnifications. Looking at Fig. 2a-b, some bright-colored regions are apparent, in which the added titanium reacted with the hBN matrix. According to the EDS point analysis from a point in the dark area (Fig. 2b), it is clear that the concentration of titanium is low in them, suggesting an almost pure hBN in such zones. By contrast, what happened within the bright colored areas is in agreement with the XRD analysis in Fig. 1. The high-magnification FESEM micrograph from such a zone (Fig. 2c) comprises some phases with different contrasts. On account of the EDS point analysis from the dark phase (not shown here), it was apparent that the dark ingredient in this image is associated with the remaining hBN. Additionally, the EDS result (Fig. 2d) related to the bright color compound (point B) revealed its consistency with the in-situ formed phase of TiB2. Nevertheless, no distinct boundary was observable between these two phases, namely hBN and TiB 2 . The EDS point  In terms of remaining porosity, it can be seen in both Fig. 2c, and Fig. 3 that a considerable amount of porosity remained within the microstructure of the hBN-20 wt% Ti sample. Excluding the observable pores between the different available phases, indeed, some free space is available amongst the hBN platelets. Comparing the morphology of the starting hBN powder with that in Fig. 2c, it can be concluded that a new lamellar structure was formed over the sintering thanks to the nucleation and growth of the new hBN platelets. As such plates grow in one direction and there is no strong bonding between the grown platelets, some sub-micron porosity can be found in their intervals. Consequently, reaching a precise value of relative density for hBN-based materials would be challenging due to the difficulty of water penetration into the interval space. Fig. 4 presents FESEM fractographs at different magnifications. Looking at these fractographs, the planar growth of new hBN platelets is clear. In addition, no distinct grains can be seen in these images, given the kind of recrystallization of the hBN particles. Nevertheless, some bright color phases are detectable in the fractographs, which are   associated with the in-situ synthesized TiB 2 and TiN ingredients.

Microstructural study
Although quite a few pores can be seen in these images, it should be taken into account that the majority of these free spaces are related to the pulled-out hBN plates over fracturing. Besides, Fig. 5 indicates a high-magnification fractography of the reaction zone in the hBN-20 wt% Ti specimen, together with the relevant EDS line results. As is apparent, each point possesses a unique chemical composition that could be due to the diffusion phenomenon, as explained earlier. It is also worth mentioning that the high reactivity of Ti with the hBN matrix prohibited the added Ti to act as a binder. In other words, as can be observed in the FESEM images, the incorporated Ti reacted with adjacent hBN matrix before it could be melted and fill the free spaces.

4.
The hBN-Ti system was studied in this investigation by incorporating 20 wt% Ti as an additive into the hBN matrix. The sample was fabricated at 1900 °C via the SPS method, characterized by XRD, EDS, and FESEM. The composite reached a relative density of ~95%. The XRD spectrum revealed the total consumption of Ti during the SPS process, generating in-situ TiB 2 as well as non-stoichiometric TiN 0.9 . The microscopical images indicated the nucleation and growth of the new hBN platelets from the initial fine particles. In addition, some boron and nitrogen atoms could leave the TiB 2 and TiN structures, respectively, diffusing into the other phases.