Synthesis, Atomic Structures and Properties of Boron Nitride Nanotubes

and electronic states, total energy calculations were carried out by molecu‐ lar mechanics and molecular orbital calculations. These studies will give us a guideline for the synthesis of the BN nanotubes, which are expected for the future nanoscale devices. The purpose of the present work is to synthesis metal-filled BN nanotube and various BN nanomaterials and to investigate the morphology of Fe-filled BN nanotube by HREM, high-an‐ gle annular dark-field scanning transmission electron microscopy (HAADF-STEM), electron diffraction and energy dispersive X-ray spectroscopy (EDX). It is possible to use HAADFSTEM to detect single heavy atoms on alight support. Scattering is caused by the nucleus and follows roughly a Z 2 dependence. Fe-filled BN nanotubes could be observed by performing centrifugation. It is considered that centrifugation is effective in collecting Fe-filled BN nano‐ tube because density of Fe is higher than that of BN nanomaterials. Formation mechanism of Fe-filled BN nanotube was proposed based on these results.


Introduction
Since the development of boron nitride (BN) nanotubes (Chopra et al. 1995), various types of BN nanostructured materials have been reported because of the great potential for using materials with low dimensions in an isolated environment. Many studies have been reported on BN nanomaterials and single crystals such as nanotubes (Golberg et al. 2000, Mickelson et al. 2003, bundled tubes, nanocorns, nanohorns, nanocapsules, nanoparticles, BN clusters, and BN metallofullerenes, which are expected to be useful as electronic devices, field-effect transistors (Radosavljevi et al. 2003), high heat-resistant semiconductors, insulator lubricants, nanowires (Tang et al. 2002), magnetic nanoparticles, gas storage materials (Lim et al. 2007), and optoelectronic applications including ultraviolet light emitters. Theoretical calculations on BN nanomaterials such as nanotubes (Rubio et al. 1994), cluster-included nanotubes, BN clusters, BN metallofullerenes, cluster solids, nanohorns, and hydrogen storage have also been carried out for prediction of the properties. By controlling the size, layer numbers, helicity, compositions, and included clusters, these cluster-included BN nanocage structures with bandgap energy of ~6 eV (Watanabe et al. 2004) and nonmagnetism are expected to show various electronic, optical, and magnetic properties as shown in Fig. 1. The differences between BN and carbon nanomaterials (Oku et al. 2009) are summarized as shown in Table 1.
The present review shows BN nanotubes synthesized by arc melting and thermal annealing methods. They were characterized by high-resolution electron microscopy (HREM), and their properties were investigated and discussed. In order to confirm the atomic structures and to investigate stabilities and electronic states, total energy calculations were carried out by molecular mechanics and molecular orbital calculations. These studies will give us a guideline for the synthesis of the BN nanotubes, which are expected for the future nanoscale devices.

Synthesis of BN nanotubes 2.1. Arc-melting of boride powders
The purpose of the present work was to prepare the BN nanotubes by arc-melting YB 6 powder in nitrogen and argon gas atmosphere. Yttrium (Y) had been reported to show excellent catalytic properties for producing single-walled carbon nanotubes (Saito et al. 1995). In the present work, YB 6 was selected to take advantage of this excellent catalytic effect (Narita & Oku, 2003). It is not necessary to prepare the boride-rod if the YB 6 powder is used. To understand the formation mechanism of BN nanotubes, HREM and electron dispersive X-ray spectroscopy (EDX) were carried out.
The YB 6 powder (4.0 g, 99.6%, Kojundo Chemical Lab. Co., Ltd) was set on a copper mold in an electric-arc furnace, which was evacuated down to 1.0×10 -3 Pa. After introducing a mixed gas of Ar (0.025 MPa) and N 2 (0.025 MPa), arc-melting was applied to the samples at an accelerating voltage of 200 V and an arc current of 125 A for 10 s (. Arc-melting was performed with a vacuum arcmelting furnace (NEV-AD03, Nissin Engineering Co., Ltd). Samples for HREM observation were prepared by dispersing the materials on holey carbon grids. HREM observation was performed with a 300 kV electron microscope (JEM-3000F). To confirm the formation of BN fullerene materials, EDX analysis was performed by the EDAX system. Low magnification image of BN nanotubes produced from YB 6 powder by arc-melting is shown in Fig. 2(a). In Fig. 2(a), length and width of the multi-wall BN nanotubes are in the range of 4-6 mm and 4-10 nm, respectively. An electron diffraction pattern of BN nanotubes with YBx nanoparticles indicate the existence of BN and YB 2 , as shown in Fig. 2(b). In Fig.  2(a), {002} and {200} reflections of YB 2 are observed. Figure 2(c) is an EDX spectrum of BN nanotubes, and strong peak of boron, nitrogen, and Y are observed. Weak peak of copper is due to the HREM grid. The EDX results showed the composition ratio of the BN nanotubes was B/N = 1.1:1. A HREM image of a multi-walled BN nanotube is shown in Fig. 2(d).
A HREM image of BN nanotube in Fig. 3(a) shows that the BN nanotube has asymmetry layer-arrangements. The layer interval on one side of the tube is 0.34 nm. Other side is in the range of 0.34-0.70 nm, which is larger than the {002} of ordinary hexagonal BN (0.34 nm). {100} planes of YB 2 are observed after the formation of BN nanotubes at the end of it, as shown in Fig. 3(b). Amorphous B with opened-tip BN nanotube is also formed at the same time by arc-melting YB 6 powder, as shown in Fig. 3(c). A novel BN nanotube is shown in Fig. 3(d). A wavy BN layer is formed into BN nanotube by most internal BN layers.      I  2/II  3  4  5  6  7  8  9  10  11  12 13/III 14/IV 15/V 16/VI 17/VII 18/VIII  H  He  1 Li    In the present work, yttrium worked as a good catalytic element to produce BN nanotubes. Catalytic metals for the formation of BN nanotubes, nanocapsules, and nanocages, which were confirmed by experiments on arc method, are summarized in Fig. 4  In the present work, the Y worked as a good catalytic element to produce BN nanotubes. Schematic illustration of the formation mechanism of BN nanotubes is shown in Fig. 6. First, ion and radical gas that consist of Y, B, and N elements would be produced by arc melting. This ion gas would be cooled by collision with Ar and N 2 gas. In this process, Y and B ions form particles of Y + B compound, which are semi-liquid state. Since B atoms become supersaturated on cooling, Y + B particles separate out B atoms on the surface. As a result, Y + B particles are covered with amorphous B. Some amorphous B would be separated from the surface of Y + B particles. BN nanolayers are formed between separated amorphous B and surface of Y + B particle. N that is necessary to form BN nanotube is provided from environmental gas. Also, B of Y + B particle would be used to form BN nanotube, because the YB 6-x compound is thermodynamically more stable than YB 6 . In the present work, BN nanotubes with YB 2 particles are formed. Closed or opened tips of BN nanotubes would be formed by cooling rate. If enough time is not given to the formation of BN nanotubes, amorphous B with opened-tip BN nanotubes would be formed, as shown in Fig. 3

Mass production of BN nanotubes
BN nanotubes have been synthesized by arc-discharge method, as decribed in the previous section. However, the arc-discharge method is not suitable for mass production because of limitation of the plasma area, and it is difficult to control nanotube size and the number of BN layers. The purpose is to synthesize BN nanotubes by ordinary thermal annealing, and to investigate the nanostructures. An Ellingham diagram of nitride metals for N 2 gas per mol was thermodynamically calculated by HSC Chemistry (Outokumpu Research Oy. Poli, Finland) software as shown in Fig. 7.
Fe 4 N particles would be reduced to α-Fe completely by annealing with boron, because boron reacted with nitrogen more easily compared to Fe. Similarly, several nitrides would be reduced to pure metals by reaction with boron. In the present work, Fe was selected for the BN nanotube formation, and a mixture powder of Fe 4 N/B was used for the synthesis ).   and B (99%, KCL) were about 5, 850, 50, and 45 lm, respectively. After Fe/B and Fe4N/B (Weight ratio [WR] = 1:1, respectively) were well mixed in a triturator, the samples were set on an alumina boat and annealed in the furnace. The furnace was programmed to heat at 6 °C /min from a room temperature to 450, 700, and 1000 °C and hold for 1-24 h, and then cooled at 3 °C /min to a room temperature. Nitrogen pressure was 0.10 MPa, and its gas flow was 100 sccm.  were well mixed in a triturator, the samples were set on an alumina boat and annealed in the furnace. The furnace was programmed to heat at 6 °C /min from a room temperature to 450, 700, and 1000 °C and hold for 1-24 h, and then cooled at 3 °C /min to a room temperature. Nitrogen pressure was 0.10 MPa, and its gas flow was 100 sccm.
X-ray diffraction patterns of samples are shown in Fig. 9(a). Diffraction peaks of hexagonal BN and a-Fe were observed for each sample except for a sample synthesized from boron powder. Diffraction peaks of B 2 O 3 were also observed for each sample except for a sample synthesized from Fe 4 N/B powder. X-ray diffraction patterns of samples synthesized from Fe 4 N/B were investigated at various temperatures and time. In Fig. 9(b), Fe 4 N was reduced to Fe by boron at temperatures in the range of 450-700 °C, and BN was obtained at 1000 °C.

Physical and Chemical Properties of Carbon Nanotubes
Phases of the samples were determined by X-ray diffraction, which showed peaks of hexagonal BN and α-Fe. Large amounts of BN nanotubes were produced, and Fig. 10(a) is a typical transmission electron microscope (TEM) image of the samples. BN nanohorn and nanotubes are observed, and lengths and widths of BN nanotubes were approximately 1-10 mm and 40-200 nm, respectively. A Fe nanoparticle is observed at the root area of a BN nanohorn. A nanotube shown by an arrow is a Fe-filled BN nanotube. Figure 10(b) is a TEM image of BN nanotube with a Fe nanoparticle, and the length is more than 2 μm. Figure  10(c) is a high magnification image of Fig. 10(b), and the BN nanotube has a bamboo-type structure, as indicated by an arrow. BN nanocoil was also produced, as shown Fig. 10(d), and a Fe nanoparticle is observed as indicated by an arrow. In the case of using magnetic materials as the catalysis metal for BN nanotubes, the magnetic nanoparticles move or rotate with the change of magnetic field, which arises from a coil heater, in the process of reaction. Therefore, it is considered that BN nanocoils were produced. High WR of Fe 4 N would be suitable for synthesis of BN nanocoils because the frequency of moving is high with increasing of the amount of magnetic nanoparticles. Bamboo-type BN nanotubes were also observed, as shown in Fig. 10(e) and 10(f). Nanoparticles were observed at the root of the nanotubes, which would be closely related with BN nanotube growth. A small amount of nanocrystalline Fe 2 B compounds were observed at the tip of the BN nanotube ( Fig. 12). Chemical formulas that Fe 4 N reacts with B, and generates Fe and BN in the experiments can be proposed as follows: Fe 2 B and dissolution of boron were obtained, and BN was produced in the reaction expressed as eq. (1) because Fe 2 B is thermodynamically more stable than Fe 4 N. Although the Fe 2 B is stable to 1389 °C, the Gibbs-Thompson effect shown that the melting occurs at a significantly lower temperature compared to values in the standard phase diagram. Therefore, fluid-like Fe 2 B can be attained more easily. In the next process, the reaction expressed as eq.
(2) would take place.  Figure 11. Low magnification images of (a) BN nanotubes with bamboo-structures and (b) iron nanoparticle at a tip of nanotube. Enlarged images of (c) a tip and (d) an interface between the Fe nanoparticle and the nanotube.  Gibb's energy on each formula is calculated as -89:4 and -23:2 kcal for the formulas (1) and (2)

Purification of BN nanotubes
Selective synthesis and purification methods for BN nanotubes are required to use them as devices, and an efficient method for purification of BN nanomaterials is required. The key steps in purification of BN nanomaterials in the present work would be HCl, HNO 3 and pyridine treatment . As-produced soot synthesized from Fe 4 N/B via the above method was purified by the following steps. The as-produced soot were poured in 4 M HCl solution and stirred for 4 h at a room temperature. The green color of the solution provides an indication of the dissolution of Fe ions. After HCl treatment, the samples were poured in 1 M HNO 3 solution and stirred for 30 h at 50 °C. The yellow color of the solution provides an indication of the dissolution of boron. After both acid treatment, the solution was filtered and rinsed with deionized water until the pH of the filtrate became neutral and dried. Then, the samples were poured in pyridine to eliminate bulk BN, and high purity BN nanotubes with a cup-stacked structure were obtained by collecting supernatant.
X-ray diffraction patterns in a purification process are shown in Fig. 13(a). Diffraction peaks of hexagonal BN, boron and α-Fe are observed for the sample at annealed at 1000°C for 1 h as shown in Fig. 13(a). It is found that Fe was removed after HCl treatment, and boron was removed after HNO 3 treatment. After pyridine treatment, a strong peak of BN was obtained as shown Fig. 13(a). Figures 13(b) show a TEM image of the sample, and there is no obvious change of the structure during the purification process, and BN nanotubes with small sizes were obtained after pyridine treatment. It is believed that bulk Physical and Chemical Properties of Carbon Nanotubes 132 size of BN was eliminated and high purity BN nanotubes were obtained by pyridine treatment. Purification of BN nanotubes were carried out by HCl, HNO 3 and pyridine treatment to remove non-BN nanotubes such as metal catalysts, boron oxides and unreacted boron.

Nanotube growth from iron-evaporated boron
The purpose is to synthesize BN nanotubes by a normal thermal annealing method. To synthesize BN nanotubes, a Fe thin film was selected and used as a catalyst for nanotube growth in the present work. Boron (B) powders with a particle size of 45 μm (99%, Kojundo Chemical Laboratory) were used as starting materials. B powder was pressed at 100 kg mm −2 into pellets with the size of 4 mm height and 15 mm in diameter. Fe with a thickness of ca. 10 nm was evaporated on the compact at ∼10 −6 torr, and the Fe would have an island structure. The samples were set on an alumina boat and annealed in a nitrogen atmosphere. The furnace was programmed to heat at 6 °C/min from a room temperature to 1000 °C and hold for 1 h, and then cooled at 3 °C/min to a room temperature. N 2 gas pressure was 0.10 MPa, and its gas flow was 100 sccm.
SEM image of surface of the Fe-evaporated B compact after annealing is shown in Fig. 14(a). Agglomerated BN nanotubes with diameters in the range of 10-20 nm are observed, and they have a network-like structure. Fig. 14(b) is a TEM image of BN nanotubes which were removed from the pellet. Diameters and lengths of BN nanotubes are in the range of 10-20 nm and 100-500 nm, respectively, and the diameters agree well with those of SEM images in Fig. 14(a). One of the typical BN nanotubes is shown in Fig. 14(c), and a nanotube axis is indicated by z. Fig. 14(d) is a Fourier filtered HREM image of center of the same BN nanotube in Fig. 14(c), and hexagonal net planes of BN nanotube are observed clearly in the image of Fig. 14(d). A hexagonal BN ring is shown in Fig. 14(d), and the BN has a zigzag-type structure, as shown in Fig. 14(e).
Growth of carbon nanotubes was explained as a model of vapor-liquid-solid (VLS) mechanism [19]. In this model, hydrocarbon such as methane is resolved in catalyst metal nanoparticles. Supersaturated solid solution of carbon in catalyst metal was precipitated as carbon nanotubes. BN nanotube growth might be explained in a similar model. Schematic illustration of growth mechanism of BN nanotubes was proposed as shown in Fig.   15. Supersaturated solid solution of B in Fe nanoparticles was formed and reacted with N 2 gas. BN nanotubes grow from these sites, and the diameter of nanotubes depends on the particle size. Fe nanoparticles are easy to be separated from BN because Fe begins to react with BN from 1350 °C, and BN nanotubes would grow as shown in Fig. 15

Chiralities of BN nanotubes
A low magnification TEM image of BN nanotubes produced from YB 6 /Ni powder is shown in Fig. 16(a) ). The lengths and diameters of BN nanotubes are ~5 μm and 3-50 nm, respectively. Fig. 16(b) is an EELS spectrum of BN nanomaterials including BN nanotubes. Two distinct absorption features are observed at 188 and 401 eV, which correspond to boron Kedge and nitrogen K-edge onsets, respectively. The fine structure of boron in the EELS spectrum shows the hexagonal bonding between boron and nitrogen, which is indicated by presence of a sharp π* peak and the shape of the σ* peak. The EELS spectrum also shows the weak σ* peaks of B and N, which indicate the spherical structure of BN nanomaterials.
A HREM image of a B 36 N 36 cluster inside a BN nanotube is shown in Fig. 16(c). The BN nanotube has a multiwalled structure, and a diameter of the most inner tube is 1.75 nm. An atomic structure model of the center of Fig. 16(c) is shown in Fig. 16(d). Diameter and chirality of the BN nanotube are 1.747 nm and (22, 0), respectively. This kind of peapod-type selforganized structure would be useful for the nanoscale devices. Another HREM image of BN nanotubes with a bundled structure is shown in Fig. 16(e), and an atomic structure model observed from three different directions is shown in Fig. 16(f). There are some spaces among the BN nanotubes, and the space would be useful for gas storage such as hydrogen. Figure 17(a) is a HREM image of a quadruple-walled BN nanotube. In the present work, all HREM images were taken close to the Scherzer defocus (Δf S = −41.2 nm), which is an optimum defocus value of electron microscope, in order to investigate the atomic structures in detail. HREM observations and electron diffraction analysis on BN nanotubes have been reported, and direct observations of nanotube chirality were tried in the present work. An enlarged HREM image is shown in Fig. 17(b), which indicates lattice fringes in the BN nanotubes.
A filtered Fourier transform of Fig. 17(b) showed that this nanotube had a zigzag-type structure as shown in Fig. 17(c) (Oku 2011). A HREM image with clear contrast processed after Fourier noise filtering is shown in Fig. 8d. The intervals of the bright and dark dots are 0.14 nm, which corresponds to the structure of h-BN rings, as shown in Fig. 17(e). Layer intervals of each tube are 0.35 nm, as shown in Fig. 17(f). Diameters of each nanotube are 2.8, 3.5, 4.2, and 4.9 nm from the inside to outside.
Another HREM image of BN nanotube produced from YB 6 powder is shown in Fig. 18(a). Width of the multiwalled BN nanotube is 8.5 nm. The BN nanotube consists of nine layers and has asymmetry layer arrangements. Layer distances are in the range of 0.34-0.51 nm, which is larger than that of {002} of ordinary h-BN (0.34 nm). Diameters of the first and second internal nanotubes are 1.7 nm and 2.6 nm, respectively. Hexagonal net planes of BN nanotube are observed in an enlarged image of Fig. 18(b). Figure 18(c) is a filtered Fourier transform of Fig. 18(b), which indicates 002 and 100 reflections of BN structure. Inverse Fourier transform of Fig. 18(c) is shown in Fig. 18(d), which indicates the lattice fringes of hexagonal networks clearly. A h-BN ring is shown in Fig. 18(d), and the BN has an armchair-type structure.    Atomic structure models were proposed from observed diameters of BN nanotubes, which were based on layer intervals of 0.34-0.35 nm. The chirality of ( n, m ) is derived from the equation The d t means a diameter of BN nanotube with nm scale, and the a B-N corresponds to the nearest distance of boron and nitrogen atoms. For the BN nanotubes, the value of a B-N is 0.144 nm. When a BN nanotube has a zigzag structure, the value of m is zero.
Atomic structure models were proposed from observed diameters of BN nanotu which were based on layer intervals of 0.34-0.35 nm. The chirality of ( n , m ) is derived the equation The d t means a diameter of BN nanotube with nm scale, and the a B-N corresponds to nearest distance of boron and nitrogen atoms. For the BN nanotubes, the value of a B-0.144 nm. When a BN nanotube has a zigzag structure, the value of m is zero. Physical and Chemical Properties of Carbon Nanotubes Figure 19(a) shows a proposed structure model of the quadruple-walled BN nanotube. Chiralities of each zigzag BN nanotube are (35, 0), (44, 0), (53, 0), and (62, 0) from the inside to outside. These chiralities were derived from (3). The arrangement of boron and nitrogen atoms was reversed at each layer, as boron atoms exist just above the nitrogen atoms while maintaining the layer intervals of 0.35 nm. Calculated images of the proposed model as a function of defocus values are shown in Fig. 19(b). Contrast of hexagonal rings was clearly imaged at the defocus values in the range of −40 to −50 nm, and these simulated images agree well with the observed HREM image of Fig. 17(d).
A proposed structure model of double-walled BN nanotube corresponding to Fig. 18 is shown in Fig. 19(c). Chiralities of the BN nanotube are (13,13) and (19,19) for the first and second layers, respectively. Layer intervals of lattice fringes of {002} planes are accorded with observed ones in Fig. 18(a). Based on the projected structure model, image calculations were carried out for various defocus values, as shown in Fig. 19(d) and a HREM image calculated at −40 nm agrees well with the experimental data of Fig. 18(d).  (13,13) and (19,19) for the first and second layers, respectively. (d) Calculated images of the proposed model (c).

BN nanotubes with cup-stacked structures
Figure 20(a) shows TEM image of BN nanotubes with a cup-stacked structure after purification process (Oku et al. 2007). Diameters and lengths of the BN nanotubes are in the range of 40-100 nm and 5-10 μm, respectively. Fe nanoparticles and bulk BN was eliminated during the process. An enlarged image of one of the BN nanotubes is shown in Fig. 20(b), which shows a cup-stacked structure as indicated by lines of BN {002}. Figure 20(c) is an electron diffraction pattern of Fig. 20(b). 002 reflections of BN are splitting in Fig. 20(c), which indicates that the BN nanotube has a cup-stacked structure and the cone angle between the BN layers at both nanotube walls is ~20°. Most of BN nanotubes (~90%) have this cup-stacked structure with cone angle of ~20°, and normal structures with a cone angle of 0° were sometimes observed (~10%). An optical absorption spectrum of BN nanotubes is shown in Fig.  20(d). In Fig. 20(d), a strong peak is observed at 4.8 eV, which would correspond to the energy gap of BN nanotubes. A broad, weak peak is also observed around 3.4 eV, which is considered to be impurity level (oxygen or hydrogen) of the BN layers. Comparable data (4.5-5.8 eV) were reported for other optical measurements (Lauret et al. 2005).
A HREM image of edge of the nanotube side wall in Fig. 20(b) is shown in Fig. 21(a), and a cup-stacked structure was observed. Edge structures are observed as indicated by arrows, and the BN {002} planes are inclined compared to nanotube axis (z-axis). Figure 21(b) is a processed HREM image after Fourier filtering of nanotube center of Fig. 21(b), and hexagonal arrangements of white dots are observed, which would correspond to BN six-membered rings. From these observations, a structure model for BN cup-structure was proposed, which consists only of h-BN rings, as shown in Fig. 21(c) and (d).
Based on the structure model of a four-layered cup-stacked B 2240 N 2240 nanotube, an image calculation was carried out as shown in Fig. 21(e). Enlarged calculated HREM images of the edge and the center of the BN nanotube in Fig. 21(e) are shown in Fig. 21(f), 21(g), respectively. These calculated images agree with the experimental data of Fig. 21(a), 21(b), respectively.
As shown in Fig. 20(c), BN layers are often inclined compared to nanotube axis, which are called cup-stacked nanotubes. A HREM image and Fourier filtered image of nanotube wall of bamboo-type BN nanotube with cup-stacked structures (WR = 1:1) is shown in Fig. 22(a) and 22(b), respectively. The nanotube axis is indicated by z-axis. BN {002} layers are inclined compared to the nanotube axis, and the cone angle between the BN layers at both nanotube walls is ~36° (Nishiwaki et al. 2005). An enlarged image of nanotube center is shown in Fig.  22(c), and a HREM image with clear contrast was processed after Fourier noise filtering as shown in Fig. 22(d), which shows hexagonal arrangements of white dots.
A structure model for B 494 N 494 cup-layer was proposed, which consists only of hexagonal BN rings. A structure model and calculated HREM images of four-fold walled B 1976 N 1976 nanotube with a cup-stacked structure are shown in Fig. 22(e) and 22(f), respectively. The calculated images (Fig. 22(f)) at defocus values of 40 and 50 nm have similar contrast of the HREM images in Fig. 22(b) and 22(d).
In order to investigate the stability of the cup-stacked structure, four types of nanotubes are considered, as shown in Fig. 23. Atomic structure models of double-walled BN nanotubes with zigzag-type and armchair-type structures, respectively, are shown in Fig. 23(a) and 23(b). Atomic structure models of four-layered, cup-stacked BN nanotubes with different cone angles are shown in Fig. 23(c) and 23(d). The values of these structures were summarized as in Tables 2 and 3. Total energies of these four-type structures indicates that BN multilayered nanotubes with and without a cup-stacked structure would be stabilized by stacking h-BN networks.    Distance between BN layers of nanotubes with a cup-stacked structure in a HREM image was found to be ~0.35 nm, and the basic structure model was constructed based on this observation. Geometry optimizations at molecular mechanics level result in the interlayer distances of ~0.38 nm. Comparing the empirical total energies of all the considered structures, a cup-stacked structure (B 2240 N 2240 ) with cone angle of 20° was found to be the lowest in energy, which indicates the high stability of this structure.
The BN nanotubes with cup-stacked structures in the present work would also be one of the candidates for atomic and gas storage, as well as carbon nanotubes. Cone angles of BN cupstacks were measured to be ~36°, which agreed well with that of the model in Fig. 22(e) (38°). Cone angles of carbon nanotubes with a cup-stacked structure were reported to be in the range of 45-80° (Endo et al. 2003). The cause of the different cone angles of the present cup-stacked BN nanotubes would be due to the different stacking of BN layers along c-axis (B-N-B-N...) from carbon layers. The cone angles might also depend on the shape of catalysis particles, as shown in Fig. 11(b). .

STM observation of BN nanotube
Although the network structure of carbon nanotubes has already been observed by scanning tunneling microscopy (STM) (Wilder et al. 1998), only few works on the STM observation of the hexagonal plane of BN nanotubes have been reported because of the insulating behavior. The STM image of BN nanotubes on highly oriented, pyrolytic graphite (HOPG) is shown in Fig. 24(a) . Three BN nanotubes are observed in the image, and the smallest one is selected for enlarged observation and electronic measurements. The nanotube axis is indicated as the z -axis. An enlarged image of the surface of the BN nanotube is shown in Fig. 24(b). The surface of the BN nanotubes is indicated by arrows. A lattice image of the BN nanotubes is observed, and an enlarged STM image of the BN nanotubes is shown in Fig. 24(c). Hexagonal arrangements of dark dots are observed, which correspond to the size of the sixmembered rings of BN. Current-voltage (I-V) measurements were also carried out for the BN nanotubes, as shown in Fig. 24(d). The I-V curve indicates an onset voltage at 5.0 V, which agreed with optical measurement of Fig. 20

Metal nanowires encapsulated in BN nanotubes
Several studies have been reported on metal-filled BN nanomaterials. Nanowires constructed from magnetic materials, especially Fe, Co and some Fe-based alloys are of interest, because they are likely to be used in nanoelectronics devices, magnetic recording media and biological sensors. However, the oxidation-and corrosionresistances of surface are weak point of the metallic nanowires. BN nanocables are of potential use for nanoscale electronic devices and nanostructured ceramic materials because of providing good stability at high temperatures with high electronic insulation in air. Therefore, metal-filled BN nanomaterials would have significant advantages for technological application. Although it is reported that Fe-filled BN nanotube could be achieved (Golberg et al. 2003), they still have some problems such as little production and low yield because it is difficult to exist in directly fabricating BN nanocable with metal cores to the poor wetting property of BN to metal. The purpose of the present work is to synthesis metal-filled BN nanotube and various BN nanomaterials and to investigate the morphology of Fe-filled BN nanotube by HREM, high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), electron diffraction and energy dispersive X-ray spectroscopy (EDX). It is possible to use HAADF-STEM to detect single heavy atoms on alight support. Scattering is caused by the nucleus and follows roughly a Z 2 dependence. Fe-filled BN nanotubes could be observed by performing centrifugation. It is considered that centrifugation is effective in collecting Fe-filled BN nanotube because density of Fe is higher than that of BN nanomaterials. Formation mechanism of Fe-filled BN nanotube was proposed based on these results.
Fe 4 N (99%, Kojundo Chemical Laboratory (KCL) Co. Ltd., Saitama, Japan) and boron (B) powders (99%, KCL) were used as raw materials. Their particle sizes were about 50 and 45 mm, respectively. After the Fe4N and B (weight ratio WR = 1:1) were mixed by a triturator, the samples were set on an alumina boat and annealed in the furnace. The furnace was programmed to heat at 6 °C/min from ambient to 1000 °C and hold for 1-5 h and then cooled at 3 °C /min to ambient temperature. Nitrogen pressure was 0.10 MPa, and its gas flow was 100 sccm. Asproduced soot synthesized via the above method was centrifuged at 8000 rpm for 2 min, and supernatant liquid is removed. The remaining sediments were collected and observed.   Fig. 26(e). Several edge-on dislocations are observed as indicated by arrows, which would be due to lattice distortion produced during Fe-filled nanotube growth. This lattice distortion is also observed as expansion in the electron diffraction pattern of Fig. 3(f), as indicated by arrows. These unique structures would be suitable materials for nanoelectronics devices, magnetic recording media and biological sensors with excellent protection against oxidation and wear.

Physical and Chemical Properties of Carbon Nanotubes 150
A HREM image of a BN nanotube synthesized from YB 6 powder is also shown in Fig. 27(a), which was taken nearly at Scherzer defocus. Number of BN {002} layers is 12, and lattice fringes are observed in the BN nanotube. A filtered Fourier transform of Fig. 27(a) is shown in Fig. 27(b). Spots of BN 002 are observed as bright spots. In addition, reflections corresponding to the yttrium structure are observed and indexed with the incident electron beam along the [101] direction. Figure 27(c) is an inverse Fourier transform of Fig. 27(b), and BN{002} layers are clearly observed in the image. An enlarged image of Fig. 27(c) is shown in Fig. 27(d), which indicates lattice fringes at the center of the BN nanotube [91]. Lattice parameters of yttrium with a hexagonal structure, as determined by X-ray diffraction analysis, were a = 0.36474 nm and c = 0.57306 nm, which agrees well with the present lattice fringes ). Dark contrast corresponds to yttrium atom pairs, as indicated in Fig. 27(d).
Based on the observations, an atomic structure model of yttrium along [101] was constructed as shown in Fig. 27(e), which indicates the yttrium atom pairs. Figure 27(f) is a calculated diffraction pattern of Fig. 27(e), and tense well with the observed Fourier transform of Fig. 27(b). Since YB 6 powders formed BN nanotubes in the present work, boron atoms were consumed preferentially. As a result, yttrium element would remain in the BN nanotube as a nanowire. These BN nanotubes with metal nanowires would be interesting nanomaterials for nanocables.

Conclusion
BN nanotubes with zigzag-, armchair-type and cup-stacked structures were synthesized and investigated by HREM, image simulation and total energy calculation. Hexagonal networks of BN nanotubes were directly observed by HREM in atomic scale, and chiralities of the BN nanotubes were directly determined from HREM images. Atomic structure models for quadruple-and double-walled nanotubes were proposed, and simulated images based on these models agreed well with experimental HREM images. Molecular mechanics calculations showed good stability of a zigzag-type structure compared to the armchair-type structure, which agreed well with the experimental data of disordered armchair-type BN nanotubes. BN nanotubes encapsulating a B 36 N 36 cluster, and yttrium and Fe nanowires were also produced and confirmed by HREM and diffraction calculation.