Synthesis of Inorganic Nanotubes

Recently, a few techniques for the synthesis of large amounts of WS2 and MoS2 multiwall inorganic nanotubes have been described [8,9,18,19,20,21,22,23]. Each of these techniques is very different from the others and produces nanotube material of somewhat different characteristics. This fact by itself indicates that the nanotubes of 2D metal-dichalcogenides are a genuine part of the phase-diagram of the respective constituents. It also suggests that, with slight changes, these or related techniques can be used for the synthesis of nanotubes from other inorganic layered compounds.

The inorganic nanotube series consisting of BxCyNz is particularly interesting, in that various stoichiometries, including BN, BC3, and BC2N have been predicted [11] and experimentally realized [24,25]. For BN, single-wall nanotubes have been synthesized and techniques now exist for the mass-production of mono-disperse double-walled BN nanotubes, including crystalline "ropes" of double-walled BN nanotubes. Multiwall BKCyNz tubes are also easily produced.

One of the earliest synthesis methods [18,19,20,21,22] of inorganic nanotubes made use of the chemical vapor transport method, which is the standard growth technique for high quality single crystals of layered metal-dichalcogenide (MX2) compounds. According to this method, a powder of MX2 (or M and X in the 1:2 ratio) is placed on the hot side of an evacuated quartz ampoule, together with a transport agent, like bromine or iodine. A temperature gradient of 20-50°C is maintained. After a few days, a single crystal of the same compound grows on the cold side of the ampoule. Accidentally, microtubes and nanotubes of MoS2 were found on the cold end of the ampoule. These preliminary studies were extended to other compounds, like WS2 (see for example, Fig. 4), and the method was optimized with respect to the production of nanotubes rather than for bulk crystals. Since the synthesis duration of this process is rather long (over two weeks time), it would be rather difficult to optimize the present synthesis method for the production of bulk amounts of such nanotubes. Whereas nanotubes produced by chemical vapor transport were found to consist of the 2H polytype [19], microtubes of diameters exceeding 1|j.m and lengths of a few hundreds of micrometers were found to prefer the 3R polytype [21] (see below).

An alternative method for the synthesis of MoS2 nanotubes has also been reported [23]. This synthesis is based on a generic deposition strategy, which has been advanced mostly through the work of Martin [26]. Nonuniform electrochemical corrosion of aluminum foil in an acidic solution produces a dense pattern of cylindrical pores, which serve as a template for the deposition of nanofilaments from a variety of materials. Thermal decomposition of a (NH4)2MoS4 precursor, which was deposited from solution at 450°C and the subsequent dissolution of the alumina membrane in a KOH solution led to the isolation of large amounts of MoS2 nanotubes. However, due to the limited stability of the alumina membrane, the annealing temperatures of the nano-

Fig. 4. Two entangled spiral WS2 nanotubes obtained by the chemical vapor transport method [19]

tubes were relatively low, and consequently their structure was imperfect. In fact, the MoS2 nanotubes appeared like bamboo-shaped hollow fibers [26].

The synthesis of a pure phase of WS2 nanotubes, 2-10|im long and with diameters in the range of 20-30 nm has been recently reported [8,9]. A typical assemblage of such nanotubes is shown at two magnifications in Fig. 5. Here, short tungsten-oxide nano-whiskers (50-300 nm long) react with H2S under mild reducing conditions. The oxide precursors are by heating a tungsten filament prepared in the presence of water vapor. One can visualize the growth process of the encapsulated nanowhisker as follows. At the first instant of the reaction (Fig. 6a), the short oxide nanowhisker reacts with H2S and forms a protective tungsten disulfide monomolecular layer, which covers the entire surface of the growing nanowhisker, excluding its tip. This WS2 monomolecular skin prohibits coalescence of the nanoparticle with neighbor-

Fig. 5. Scanning electron microscopy image (two magnifications) of a mat of WS2 nanotubes obtained from a powder of short asymmetric oxide nanoparticles by a combined reduction/sulfidization process [9]

Fig. 6. Schematic representation of the growth model of the WS2 nanotubes [9]

a be

Fig. 6. Schematic representation of the growth model of the WS2 nanotubes [9]

ing oxide nanoparticles, which therefore drastically slows their coarsening. Simultaneous condensation of (WO3)n or (WO3_a;-H2O)n clusters on the uncovered (sulphur-free) nanowhisker tip, and their immediate reduction by hydrogen gas, lead to the lowering of the volatility of these clusters and therefore to the tip growth. This concerted mechanism leads to a fast growth of the sulfide-coated oxide nanowhisker. Once the oxide source is depleted, the vapor pressure of the tungsten oxide in the gas phase decreases and the rate of tip growth slows down. This leads to the termination of the growth process, since the rate of the sulfidization of the oxide skin continues at the same pace and the exposed whisker tip becomes coated with the protective sulfide skin. This process leads to the formation of oxide nanowhiskers 2-10|j.m long, coated with an atomic layer of tungsten sulfide. After the first layer of sulfide has been formed, in an almost instantaneous process, the conversion of the oxide core into tungsten sulfide is a rather slow diffusion-controlled process. The oxide nanowhisker growth is schematically illustrated in Fig. 6b. Note that during the gradual reduction of the oxide core, the CS planes in the oxide phase rearrange themselves and approach each other [27] until a stable reduced oxide phase-W3O8 is reached [28]. This phase provides a sufficiently open structure for the sulfidization to proceed until the entire oxide core is consumed and converted into a respective sulfide. Furthermore, the highly ordered nature of the reduced oxide provides a kind of a template for a virtually dislocation-free sulfide layer growth. Further reduction of the oxide core would bring the sulfidization reaction to a halt [29]. It was indeed shown that in the absence of the sulfide skin, the oxide nanowhisker is reduced rather swiftly to a pure tungsten nanorod. Therefore, the encapsulation of the oxide nanowhisker, which tames the reduction of the core, allows for the gradual conversion of this nanoparticle into a hollow WS2 nanotube. The present model alludes to the highly synergistic nature between the reduction and sulfidization processes during the WS2 nanotube growth and the conversion of the oxide core into multiwall tungsten sulfide nanotubes. It is important to emphasize that the diameter of the nanotube is determined by the precursor diameter in this process. The number of WS2 layers increases with the duration of the reaction until the entire oxide is consumed. This permits very good control of the number of walls in the nanotube, but with the oxide core inside. The oxide whisker can be visualized as a template for the sulfide, which grows from outside inwards. This method does not lend itself to the synthesis of (hollow) single wall nanotubes, since long whiskers, a few nm thick, would not be stable.

The multiwall hollow WS2 nanotubes, which are obtained at the end of the process are quite perfect in shape, which has a favorable effect on some of their physical and electronic properties. This strategy, i.e., the preparation of nanowhiskers from a precursor and their subsequent conversion into nanotubes, is likely to become a versatile vehicle for the synthesis of pure nanotube phases from other 2D layered compounds, as well.

Another strategy for the synthesis of inorganic nanotubes is through the sol-gel process [6,29,30,31,32]. Here, a metal organic compound is dissolved together with a template-forming species in alcohol, and a template structure for the growth of nanotubes is formed. The addition of small amounts of water leads to a slow hydrolysis of the organic-metal compound, i.e., the formation of metal-oxide sol, but the template structure is retained. Upon tuning of the pH, the sol transforms into a gel, which consists of a -M-O- polymer with longer chains. In the case of the layered compound V2O5 [29,30], a sol was first prepared by mixing vanadium (V) oxide tri-isoporpoxide with hexadecylamine in ethanol and aging the solution while stirring, which resulted in the hydrolysis of the vanadium oxide. Subsequent hydrothermal treatment at 180° C led to the formation of nanotubes with the formal composition VO2.4-(C16H33NH2 )034. These nanotubes are crystalline, which is evident from their X-ray and electron diffraction patterns. Facile and (quite) reversible Li intercalation into such nanotubes has been demonstrated [31], which puts this material into the forefront of high energy density battery research.

Thus far, only metal oxide nanotubes have been synthesized by this process. Whereas crystalline nanotubes were obtained from 2D (layered) oxides, the 3D oxide compounds, like SiO2 [32] resulted in semicrystalline or amorphous nanotubes only. In principle, this kind of process could be extended to the synthesis of nanotubes from chalcogenide and halide compounds in the future.

In another procedure, carbon nanotubes were used as templates for the deposition of V2O5 nanotubes; the template was subsequently removed by burning the sample in air at 650° C [33]. This strategy can be easily adopted for the synthesis of different oxide nanotubes, as shown below.

The heating of an ammonium thiomolybdate compound with the formula (NH4)2Mo3S13 • xH2O was shown to lead to its decomposition around 673K and to the formation of a non-homogeneous MoS2 phase, which also contains elongated MoS2 particles (mackles) [34]. The mackles are ascribed to a topochemical conversion of the precursor particles, which form closed layers of MoS2 on top of an amorphous core (possibly a-MoS3).

Crystallization of amorphous MoS3 (a-MoS3) nanoparticle precursors by the application of microsecond long electrical pulses from the tip of a scanning tunneling microscope, led to the formation of composite nanoparticles with an IF-MoS2 envelope and an a-MoS3 core [35]. The MoS2 shells were found to be quite perfect in shape and fully closed. This observation is indicative of the fast kinetics of the crystallization of fullerene-like structures. A self-propogating self-limiting mechanism, has been proposed [35]. According to this model, the process is maintained by the local heating due to the exothermic nature of the chemical reaction and of the crystallization process until the MoS2 layers are completed and closed [35].

A novel room-temperature method for producing nested fullerene-like MoS2 with an a-MoS3 core using a sono-electrochemical probe has been described recently [36]. MoS2 nanotubes also occur occasionally in this product. Ultrasonically-induced reactions are attributed to the effect of cavitation, whereupon very high temperatures and pressures are obtained inside imploding gas-bubbles in liquid solutions [37]. The combination of a sono-chemical probe with electrochemical deposition has been investigated for some time now [38,39]. Generally, the decomposition of the gas molecules in the bubble and the high cooling rates lead to the production of amorphous nanomate-rials. It appears, however, that in the case of layered compounds, crystalline nanomaterials with structures related to fullerenes, are obtained by sono-chemical reactions [36]. In this case, the collapsing bubble serves as an isolated reactor and there is a strong thermodynamic driving force in favor of forming the seamless (fullerene-like) structure, rather than the amorphous or plate-like nanoparticle [35,40]. In fact, there appears to exist a few similarities between the above two methods for the preparation of IF-MoS2 [35,36]. First, both processes consist of two steps, where the a-MoS3 nanoparticles are initially prepared and are subsequently crystallized by an electrical pulse [35] in one case, and by a sonochemical pulse in the other process [36]. A compelling factor in favor of the fast kinetics of fullerene-like nanoparticle formation is that, in both processes, the envelope is complete, while the core of the nanoparticles (>20 nm) remains amorphous. Since, the transformation of the amorphous core into a crystalline structure involves a slow out-diffusion of sulfur atoms, the core of the nanoparticles is unable to crystallize during the short (ns - |os) pulses. It is likely that the sonochemical formation of IF-MoS2 nanoparticles can also be attributed to the self-propagating self-limiting process described above.

In a related study, scroll-like structures were prepared from the layered compound GaOOH by sonicating an aqueous solution of GaCl3 [41]. This study shows again the preponderance of nanoparticles with a rolled-up structure from layered compounds. It has been pointed-out [42] that in the case of non-volatile compounds, the sonochemical reaction takes place on the interfacial layer between the liquid solution and the gas bubble. It is hard to envisage that a gas bubble of such an asymmetric shape, as the GaOOH scroll-like structure, is formed in an isotropic medium. Alternatively, one can envisage that a monomolecular layer of GaOOH is formed on the bubble's envelope, which rolls into a scroll-like shape once the bubble is collapsed. As indicated in [41], rolled-up scroll-like structures were obtained by sonication of InCl3, TlCl3 and AlCl3, which demonstrates the generality of this process. Hydrolysis of this group of compounds (MCl3) results in the formation of the layered compounds MOOH, which, upon crystallization, prefer the fullerene-like structures. Nevertheless, MOCl compounds with a layered structure are also known to exist and their formation during the sonication of MCl3 solutions has not been convincingly excluded. More recently, the same group reported the formation of fullerene-like Tl2 O nanoparticles by the sonochemical reaction of TlCl3 in aqueous solution [43]. Some compounds with the formula M2O, where M is a metal atom, possess the anti-CdCl2 structure, with the anion layer sandwiched between two cation layers. Currently, the yield of the IF-Tl2O product is not very high (^10%), but purification of this phase by the selective heating of the sample to 300° C has been demonstrated. Furthermore, size and shape control of the fullerene-like particles is not easy in this case. Nonetheless, the fact that this is a room temperature process is rather promising, and future developments will hopefully permit better control of the reaction products. Perhaps most important, the versatility of the sonochemical technique is a clear-cut asset for nanoparticle synthesis. It is also to be noted that Tl2O is a rather unstable compound in bulk form, and the formation of a closed IF structure appears to render it a stable phase. NiCl2 "onions" and nanotubes prepared by sublimation of a NiCl2 powder at 950° C was reported [44]. These nanotubes have potentially interesting magnetic properties. However, their large scale synthesis has only met with partial success so far, mostly due to the hygroscopic nature of the precursor.

As discussed above briefly, semi-crystalline or amorphous nanotubes can be obtained from 3D compounds and metals, by depositing a precursor on a nanotube-template intermediately, and subsequent removing the template by, for example, calcination. Since a nanotube is a rolled-up structure of a 2D molecular sheet, there is no way that all the chemical bonds of a 3D compound will be fully satisfied on the nanotube inner and outer surface. Therefore, in this case, the nanotubes cannot form a fully crystalline structure and the nanotube surface is not going to be inert. Nonetheless, there are certain applications, like in catalysis, where such a high surface area pattern with reactive surface sites (i.e., unsaturated bonds) is highly desirable. The first report of SiO2 nanotubes [45] came serendipitously during the synthesis of spherical silica particles by the hydrolysis of tetraethylorthosilicate in a mixture of water, ammonia, ethanol and tartaric acid. More recently, nanotubes of SiO2 [46], TiO2 [5,47], Al2O3 and ZrO2 [5,48], etc. have been prepared by the self-assembly of molecular moieties on preprepared templates, like carbon nanotubes, elongated micelles, or other templates which instigate uniaxial growth. In fact, there is almost no limitation on the type of inorganic compound which can be 'molded' into this shape, using this strategy. We note in passing, that the tuneability of the carbon nanotube radii and the perfection of its structure could be important for their use as a template for the growth of inorganic nanotubes with a controlled radius. The formation of nanootubular structures can be rather important for the selective catalysis of certain reactions, where either the reaction precursor or the product must diffuse through the (inorganic) nanotube inner core.

The rational synthesis of peptide based nanotubes by the self-assembling of polypeptides into a supramolecular structure, was demonstrated. This self-organization leads to peptide nanotubes having channels 0.8 nm in diameter and a few hundred nm long [49]. The connectivity of the proteins in these nanotubes is provided by weak bonds, like hydrogen bonds. These structures benefit from the relative flexibility of the protein backbone, which does not exist in nanotubes of covalently bonded inorganic compounds.

We now discuss the synthesis of BKCyNz nanotubes and nanoparticles. The similarity between graphite and hexagonal BN suggests that some of the successful synthesis methods used for carbon nanotube production might be adapted to BKCyNz nanotube growth. This is indeed the case. A non-equilibrium plasma arc technique has been used to produce pure BN nano-tubes [25]. To avoid the possibility of carbon contamination, no graphite components are used in this synthesis. The insulating nature of bulk BN prevents the use of a pure BN electrode. Instead, a pressed rod of hexagonal BN is inserted into a hollow tungsten electrode forming a compound anode. The cathode consists of a rapidly cooled pure copper electrode. During discharge, the environmental helium gas is maintained at 650 torr and a dc current between 50A and 140 A is applied to maintain a constant potential drop of 30 V between the electrodes. The arc temperature exceeds 3700 K.

Arcing the BN/W compound electrode results in a limited amount of dark gray soot deposit on the copper cathode, in contrast to the cohesive cylindrical boule which typically grows on the cathode upon graphite arcing. High resolution TEM studies of the soot reveal numerous multiwalled nanotubes. Figure 7 shows a representative TEM image of a BN nanotube thus produced. Electron Energy-Loss Spectroscopy (EELS) performed on individual nanotubes inside the TEM confirms the B-N 1: 1 stoichiometry. Figure 7 shows that the end of the BN nanotube contains a dense particle, most likely tungsten or a tungsten-boron-nitrogen compound. In contrast to carbon nanotubes, where the capping is fullerene-like or involves pentagons and heptagons, BN tube closure by pentagon formation is suppressed in order to inhibit the formation of less favorable B-B bonds. The small metal cluster may thus simultaneously act as a growth catalyst and as an aid to capping the nanotube by relieving strain energy.

Different synthesis methods have been successfully employed to produce BN nanotubes. An arc-discharge method employing HfB2 electrodes in a nitrogen atmosphere yields BN nanotubes with a wall number ranging from many to one [50]. In this synthesis method, the Hf is apparently not incorporated into the tube itself, but rather acts as a catalyzing agent. The source of nitrogen for tube growth is from the N2 environmental gas. The ends of the BN nanotubes thus produced appear to be pure BN, with unusual geometries reflecting the bond frustration upon closure. Tantalum has also been used as the catalyzing agent in the synthesis of various nanoscale BN structures using arc-vaporization methods [51]. Pure BN nanotubes are produced, along with other nanoparticles, including onion-like spheres similar to those produced by a high intensity electron irradiation method [52]. In other studies, circumstantial evidence is found for the presence of B2N2 squares at the BN nanotube tips, as well as B3N3 hexagons in the main fabric of the nanotubes [51].

A most intriguing observation is that of BKCyNz nanotubes with segregated tube-wall stoichiometry [53]. Multiwall nanotubes containing pure-carbon walls adjacent to pure BN walls have been achieved, forming a sort of nanotube coaxial structure. In one specific tube studied carefully by EELS, the innermost three walls of the tube contained only carbon, the next six walls

Fig. 7. TEM micrograph of the end of a BN nanotube showing termination by a metal particle. (courtesy of N. G. Chopra [25])

10 nm

Fig. 7. TEM micrograph of the end of a BN nanotube showing termination by a metal particle. (courtesy of N. G. Chopra [25])

were comprised of BN, and the last five outermost walls were again pure carbon. The entire 14-walled composite nanotube was 12 nm in diameter. Similar layer segregation is obtained for onion-like coverings over nanoparti-cles (quite often the core nanoparticle is composed of the catalyst material). Related studies have also found BKCyNz nanotubes consisting of concentric cylinders of BC2N and pure carbon. A reactive laser ablation method has also been used to synthesize multi-element nanotubes containing BN [54]. The nanotubes contain a silicon carbide core followed by an amorphous silicon oxide intermediate layer; this composite nano-rod is then sheathed with BN and carbon nanotube layers, segregated in the radial direction. It has been speculated that the merging BN and carbon nanotube structures may be the basis for novel electronic device architectures.

The above-described methods for BKCyNz nanotube synthesis unfortunately result in relatively low product yields, and do not produce monodisperse materials. Recently, a method has been developed which produces virtually mono-disperse BN double-walled nanotubes in large quantity [10]. Multi-wall BN "nano-cocoons" with etchable cores have also been synthesized in large quantity [10] using this method. The synthesis employs nitrogen-free, boron-rich electrodes arced in a pure nitrogen gas environment. The electrodes incorporate 1 at.% each of nickel and cobalt. Arcing in a dynamic N2 gas environment with a pressure near 380 Torr results in an abundance of gray web-like material inside the synthesis chamber; this material contains a high percentage of double-walled BN nanotubes, as shown in Fig. 8. The

Fig. 8. High resolution TEM image of a BN nanotube with two layers. Inset: Image of a BN nanotube end. (courtesy of J. Cum-ings [10])

tubes have a narrow size distribution, centered on 2.7nm (outer diameter) with a standard deviation of 0.4 nm. The double walled BN nanotubes often aggregate as "ropes", very similar to the crystalline rope-like structures previously observed for mono-disperse single-walled carbon nanotubes [55]. Figure 9 shows BN nano-cocoons before and after their cores have been etched out with acid [10]. The nano-cocoons can be formed with a relatively uniform size distribution.

The double-wall BN structure has been observed to form in the case of HfB2 arced in nitrogen gas [50] and is claimed to be the dominant structure in laser ablation of BN with a Ni/Co catalyst carried out in a nitrogen carrier gas [56]. It has been proposed that lip-lip interactions [57] could stabilize the growth of open-ended multiwall carbon nanotubes [17]. It has been predicted that the B-N bond of a single-wall BN nanotube should naturally undergo out-of-plane buckling [58], producing a dipolar shell. Buckling (absent in carbon tubes) may thus favor double-walled BN nanotube growth.

Multi-walled nanotubes with different BxCyNz stoichiometries have also been produced in small quantity, including BC2N and BC3 nanotubes [25,59]. Importantly, both BC2N and BC3 are known to exist in bulk (i.e., layered sheet) form. The bulk materials can be synthesized [60] via the following chemical reactions:

CH3CN + BCl3 —► BC2N + 3HCl (>800 °C) 2BCl3 + C6H6 —► 2BC3 + 6HCl ( 800 °C).

Both bulk BC2N and BC3 have bright metallic luster and resemble the lay-

Fig. 9. (a) BN-coated boron nanocrystal. (b) Same as (a), after treating with nitric acid. The boron core has been chemically removed, leaving an empty BN nano-cocoon. (courtesy of J. Cumings [10])

ered structure of graphite. Resistivity measurements of the layered bulk compounds indicate that BC2N is semiconducting with an energy gap of about 0.03 eV and BC3 is semi-metallic [60]. BC2N and BC3 nanotubes have been produced by arc-synthesis methods using a compound anode formed by inserting a pressed BN rod inside a hollowed out graphite electrode. The compound anode is arced against a copper cathode in a 450 torr helium gas environment. The stoichiometry of the resulting multiwalled nanotubes is identified using EELS, but further experimental characterization is lacking. In the next section we discuss some of the theoretically predicted properties of BC2N and BC3 nanotubes.

0 0

Post a comment