Different types of reaction products were obtained depending on the pressure inside the reaction chamber (Fig. 3a–d). At a reaction chamber pressure of 2 or 4 bar, white powdery material grew from the powder container to several centimeters height (Fig. 3a, b). Transmission electron microscopy (TEM (FEI technai at the Core-facility for Bionano Materials in Gachon University)) images show that the powder has a layer-by-layer structure (Fig. 3i, j) with a lattice distance of about 0.34 nm. Electron energy-loss spectroscopy (EELS) analysis revealed that B and N atoms were present across the sample with a ratio of 1:1, suggesting that the products are h-BN particles. However, at an elevated pressures of 8 or 12 bars, the main product was strands of white fibers stretching out from the powder container to the collector surface. TEM analysis shows that the fibers are composed of nano-scale fibrous materials that have a tubular structure (Fig. 3k, l). EELS analysis confirmed that the fibrous materials are composed of B and N atoms with a ratio of ~ 1:1 (Additional file 1: Fig. S1), suggesting that BNNTs are mainly produced at high pressures. Interestingly, unlike the products from the conventional laser ablation method, elemental B nanoparticles are rarely observed in this sample. In the previous studies [19] with a solid B target (e.g., B fibers), nano-sized B particles are often observed as by-products and proposed to serve as BNNT nucleation sites (i.e., BNNT seeds). However, in the current samples, majority of the impurity particles around BNNTs are h-BN particles (Additional file 1: Fig. S2) that exhibit the similar layer-by-layer structure (Fig. 3i, j) we observed with the white powder produced under the lower pressure of 2 or 4 bar. The size of h-BN impurities in the samples also depends on the pressure and decreases as the pressure increases (Fig. 3e–h). The absence of nano-sized B particles is intriguing as they are believed to play an important role in the BNNT nucleation and growth, and thus implies that a different growth mechanism is needed to explain the BNNT growth from ammonia borane.
In the previous studies of the synthesis of BNNTs by laser ablation, B and N atoms were independently supplied using solid boron and gaseous N2 molecule, respectively [19]. Although this study aimed to synthesize BNNTs by direct supply of BN units from H3N-BH3, we also speculated that additional N radical generated by dissociation of the surrounding N2 gas might have benefited the BNNT growth because the temperature of the hot spot is as high as 4000 K. A control experiment was performed with argon to investigate the effect of the surrounding gas, and almost same results were obtained; at pressures of 2 and 4 bar (Fig. 4a–c), white powdery materials were produced while BNNT fibers were formed at elevate pressures of 8 and 12 bar (Fig. 4d–f). It is also noted that the amount of reaction products were almost same compare with the N2 cases. This is probably because N2 molecule has a triple bond with high bonding energy (944.8 kJ/mol), and thus the amount of nitrogen gases decomposed into nitrogen atoms was extremely small even at a high temperature of 4000 K. The control experiment demonstrated that the type of the surrounding gas is not critical for the BNNT synthesis, and ammonia borane itself is sufficient to supply B and N atoms at a 1:1 ratio for the BNNT growth.
Our main finding from the parametric study can be summarized as follows: (i) the morphology of the main reaction product changes from h-BN to BNNTs as the chamber pressure increases; (ii) nano-sized B droplets are not produced if ammonia borane is used as feedstock; (iii) the type of the surrounding gas is not critical for the BNNT growth as it seems not participating in the BNNT synthesis reaction. Obviously, these findings are not consistent with the growth mechanism proposed by Kim et al. (i.e., dual growth mode) for the laser-grown BNNTs, and thus implies different pathways in the BNNT nucleation and growth by ammonia borane.
In ammonia borane, N-B bond is stronger than B-H and N–H bonds. Upon thermal energy absorption, the relatively weak B-H and N–H bonds are expected to break first and release H atoms. When the temperature increases gradually from room temperature to 1500 °C, ammonia borane forms polyaminoborane (-(BH2-NH2)-n) and polyiminoborane (-(BH-NH)-n) in consecutive order through dehydrogenation process. Such intermediate species eventually turn into h-BN through further hydrogen release. However, a rapid temperature increment by laser irradiation may result in an immediate decomposition of ammonia borane into various gaseous species including BN, BH, NH, and BNH radicals, rather than bulk h-BN. In this case, BN radicals are expected to be the most dominant species among the various chemical species due to its strongest bonding energy. We also suggest that the BN radicals formed do not subsequently decompose into elemental B and N atoms because elemental boron particles were seldom observed from our samples. In the previous study [19], TEM images of BNNT synthesized often show BNNTs with B nano-droplets at their tips. However, in this study, such nano-droplets were never seen from TEM analysis (Additional file 1: Fig. S3). This argument is also supported by thermogravimetric analysis (TGA) of our samples where the mass gain by B2O3 formation is not significant during oxidation (Additional file 1: Fig. S4). The mass increment from 650 °C to 800 °C in the TGA graph is due to the oxidation of elemental boron particles while the mass gain around 900 °C might be associated with the oxidation of defective BN by-products, such as amorphous or turbostratic BNs because crystalline h-BN materials start to oxidize from 950 °C. The elemental boron content in this sample is estimated as low as 1.35 wt. %. The absence of elemental boron partially explains why the BNNT growth from ammonia borane is less sensitive to the type of the surrounding gas. Unlike the conventional process, the re-nitridation of elemental boron by N radicals or excited N2 molecules (e.g., vibrationally excited N2 (ν) or N2+ ions) are not an essential step in this case. Our study found that BNNTs seem to grow directly from BN radicals when ammonia borane is used as feedstock, but at the same time it raises an important question of how BNNTs nucleate and grow directly from BN radicals in free space.
BN radicals formed can be a major precursor for continuous of growth of both h-BN and BNNTs. The fact that the morphology of the main product is determined by the reaction pressure implies that the BNNTs growth is not dictated by chemistry but by kinetics. Since h-BN has a planar structure, BN radicals could form h-BN via lateral growth as illustrated in Fig. 5a. When the pressure of the surrounding gas increases, the concentration of BN radical decreases and consequently the lateral growth facilitated via collisions among BN radicals may be inhibited by more frequent collisions with surrounding gas molecules (Fig. 5b). The rate of lateral growth of h-BN is inversely proportional to the pressure of the surrounding gas. As the growth speed slows down, the size of the h-BN fragment decreases which is in line with our experimental observation, and also the number of dangling bonds per area increases, making them more unstable. The latter can facilitate folding or zipping of h-BN fragments in order to reduce the number of unstable dangling bonds (Fig. 5b).
At high pressures, h-BN fragments also gain enough energy by collision with surrounding gases and can overcome the strain-energy barrier required for curvature formation (Fig. 5b). This phenomenon has also been reported in the kinetics of fullerene formation [27]. Laser was applied to a rotating graphite disk and carbon species were vaporized into a stream of helium (He), cooled and partially equilibrated during the expansion. When the pressure of He stream was less than 1 atm, the number of C atoms per cluster was distributed in the range of 50 to over 90; however when the pressure increased to 10 atm, clusters of 60 carbon atoms (i.e., fullerene) were selectively obtained. When the hot ring clusters are remained in contact with high-density He, they equilibrate towards the most stable species through two- and three body collisions, which appears to be a unique cluster containing 60 atoms.27 This phenomenon can be adapted to explain the pressure effect of this study and BNNT growth from BN radicals at high pressures. At high pressures, C clusters are subjected to frequent collisions with the surrounding gases which provides sufficient energy to overcome the kinetic barrier that originates from the curvature strain. Therefore, they tend to form thermodynamically-favorable species of a spherical fullerene. However, in the case of BN, frustration of bonding occurs when a B-B bond or an N–N bond is created in the pentagonal structure [28]. The energetically-favorable curved structure is a tubular structure (i.e., nanotube) rather than a BN fullerene. Starting from a curved seed as illustrate in Fig. 5b, B and N atoms form a nanotube at high pressures by eliminating energetically-costly dangling bonds in the edges. In this work, the curved B-N structure (Fig. 5b) is proposed as the nucleus of BNNT formation (i.e., homogeneous nucleation). For the growth of BNNT, BN radicals are continuously added to the curved h-BN seed and a BNNTs grow in the axial direction. BN radicals can also participate in growing other h-BN species when the edges of the curved seed are not commensurate with formation of a tubular structure. Such by-products become main impurities of the product (Additional file 1: Fig. S2).
Results of the thermo-fluid simulation such as temperature, BN mass fraction, and velocity fields along with streamline analysis, are present in Fig. 6. As the surrounding gas pressure increases, it was predicted that the high temperature zone shrinks and the BN concentration also decreases in the reaction zone (Fig. 6b). This is a pure pressure effect caused by adding more surrounding gas. The velocity distributions (Fig. 6d) predict an up-flow formation due to the evaporation of ammonia borane and the temperature difference between the laser spot and the surroundings (i.e., buoyancy force). The flow field pattern also changes slightly with the pressure. In the case of low pressure (e.g., 2 bar), the gas velocity flowing in from the upper part is larger than that of the high pressure (e.g., 12 bar) case, thus the BN precursors or BN debris generated may re-enter the reaction zone (Additional file 1: Fig. S5) which increases the amount of BN in the reaction zone. On the other hand, in the case of 12 bar, the velocity of the gas from the upper part is reduced and thus the possibility of re-entering into the reaction zone seems relatively small (Additional file 1: Fig. S5). The simulation results support our discussion that h-BN growth may slow down at high pressures due to the reduction of BN radicals in the reaction zone. It was also observed that the direction of the up-flow is fairly consistent with the direction of BNNT fiber formation. It seems that the up-flow from the hot spot helps in-situ fiber formation by confining and directing BNNTs toward the flow direction (i.e., flow-driven alignment).
The degree of alignment of BNNT fibers was also analyzed by the Raman signal intensity (I) in VV mode (Fig. 7a). BNNT fibers show a peak near 1360 cm−1, attributed to the E2g vibrational mode in the h-BN sheet [29]. The Raman signal intensity was maximum at 0° while minimum at 90°. Previous research has shown that the intensity I in VV mode is [30]
$$I_{\theta } = \, A \, cos^{4} \theta \, + \, B,$$
(1)
0°/I90° = 2.1. These results show a degree of alignment like the I0°/I90° = 2.2 reported previously for BNNT alignment [31] The data used for calculation are average values obtained by measuring each sample at least three times. From the data, we can conclude that the well-aligned BNNT fiber has been formed from the method we report.