- Email: [email protected]
Large-scale synthesis and HRTEM analysis of single-walled B- and N-doped carbon nanotube bundles
PERGAMON
Carbon 38 (2000) 2017–2027
Large-scale synthesis and HRTEM analysis of single-walled B- and N-doped carbon nanotube bundles D. Golberg*, Y. Bando, L. Bourgeois, K. Kurashima, T. Sato National Institute for Research in Inorganic Materials, Namiki 1 -1, Tsukuba, Ibaraki 305 -0044, Japan Received 17 November 1999; accepted 21 February 2000
Abstract Bundles of B- and N-doped single-walled carbon nanotubes (SWNTs) containing up to |10 at% B and up to |2 at% N were synthesized at high yields under thermo–chemical treatment of pure C SWNT bundles and B 2 O 3 in a flowing nitrogen atmosphere. The bundles were characterized by means of high-resolution transmission electron microscopy and electron energy loss spectroscopy. The effects of synthesis temperature (1503–1773 K) and time (30–240 min) on the B and N contents and yield of the SWNT bundles were determined. The maximum yield of the B- and N-doped SWNT bundles was obtained under synthesis at 1553 K over 30 min. Atomic structure and morphology of individual SWNTs in the bundles, in particular, packing of doped SWNTs, helicity distribution, encapsulation of fullerene-like clusters, diameter and shell number variations were studied. The synthesized SWNTs in the bundles were stacked in a honeycomb array with the uniform inter-tube spacing of |0.3 nm. No preferable orientation for the graphene-like tubular shells was found, i.e. both zigzag and armchair edges were observed with approximately equal proportions. Frequently, diameter increase took place for the outer tubes in a bundle and for isolated SWNTs. C-based or BN-based fullerene-like encapsulates were observed in individual SWNTs. Carbon oxidation by the B 2 O 3 vapor and B and N substitution for C is thought to underlie the doping of C SWNTs. The substitution reaction temperature–time limits with respect to the morphological stability of B- and N-doped SWNT bundles are finally elucidated. 2000 Elsevier Science Ltd. All rights reserved. Keywords: A. Doped carbons; Carbon nanotubes; C. Transmission electron microscopy (TEM); electron energy loss spectroscopy (EELS); D. Microstructure
1. Introduction Nanotubes made of carbon [1] are expected to bring significant breakthroughs in the technology of electronic and engineering materials of the next millennium. For applications, single-walled nanotubes (SWNTs) are preferred over other nanotubular structures [2], since the number of structural defects in them, which can dramatically affect the properties, is significantly reduced. Doping of C nanotubes with B and / or N [3–8] or preparing sandwich-like structures made of C and BN layers [9] may allow to tailor nanotube electronic [10,11] and mechanical properties [12]. For instance, B-doped nanotubes were found to exclusively exhibit metallic conductivity [10] in contrast to undoped nanotubes whose properties vary between metallic and semiconducting depending on helici*Corresponding author. Fax: 181-298-51-6280. E-mail address: [email protected] (D. Golberg).
ty and diameter [13]. N-doped nanofibres were also expected to be metallic [8]. In addition, B-doping was found to improve graphitization of multi-walled carbon nanotubes (MWNTs) [4]. Also, it is known that B enhances the oxidation resistance of graphite and conventional carbon fibers [14] which is important for applications. Doping of C MWNTs and preparation of composite B–C–N MWNTs are usually achieved by arc-discharge [4] or laser ablation [6] by changing the chemical composition of starting electrodes or targets, respectively. Both processes occur in the highly non-equilibrium conditions which negatively influence the chemical and structural homogeneity of the product. An alternative way of effecting C MWNT chemical composition variations has recently been discovered: A general chemical substitution reaction during which C atoms in the nanotubular network are substituted with B or B and N atoms under nanotube oxidation by the B 2 O 3 vapor in an argon or nitrogen atmosphere [15,16]. Later on, this method has been
0008-6223 / 00 / $ – see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S0008-6223( 00 )00058-0
2018
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
successfully applied for changing chemical composition of C SWNT bundles by the present authors [17]. However, the limited yield and insufficient chemical homogeneity of the doped SWNT product are serious drawbacks which so far have made a detailed characterization of B- and Ndoped SWNT bundles difficult [17]. The present paper reports on the pioneering of large-scale synthesis of B- and N-doped C SWNT bundles by optimization of the synthesis parameters during the substitution reaction and detailed analysis of the resultant doped SWNT product by means of high-resolution transmission electron microscopy paired with electron energy loss spectroscopy.
2. Experimental Bundles of pure C SWNTs produced by Nd:YAG laser ablation either by CarboLex (USA) or JST-ICORP (Japan) were heated together with B 2 O 3 in a flowing nitrogen atmosphere at 1503–1773 K over 30–240 min. The heating was carried out in a vertical induction furnace with a susceptor made of graphite [15–17]. The B 2 O 3 powder was placed in an open sintered graphite crucible and then covered with C SWNTs. The experimental set-up is shown in Fig. 1. The N 2 gas was introduced into the chamber at the ambient pressure from the upper and lower inlets (Fig.
1) and its flows were maintained at 3 l / min (upper flow) and 0.2 l / min (lower flow) during the synthesis. The synthesis temperature was monitored using an optical pyrometer with the accuracy of 610 K. Upon completing the synthesis the product of the reaction was extracted from the crucible, milled in an agate mortar and mixed with CCl 4 . A few drops of the resultant mixture were dripped onto a standard [3 mm carboncoated-cooper grid. High-resolution transmission electron microscopy (HRTEM) was carried out by means of a field emission electron microscope JEM-3000F (JEOL) operating at 300 kV. The microscope had the following characteristics: Spherical aberration coefficient — 1.1 mm; chromatic aberration coefficient — 1.8 mm, point resolution — 0.16 nm; stability of high voltage — 4310 26 ; and tilt angle 6308. HRTEM images were taken at magnification 4310 5 on a negative film at the vicinity of the optimal defocus value of the microscope, — 56 nm. Parallel detection electron energy loss spectroscopy (Gatan 666) was performed with the aim of measuring B and N contents in the product. Normally, an electron beam focused down to [0.5–1.6 nm was placed at different points along and across the SWNT bundles for measuring chemical content variations. Computer simulated SWNT HRTEM images were calculated and displayed using the ‘MacTempas’ software.
3. Results and discussion
3.1. Effect of synthesis parameters on the B /N contents and yield of doped SWNT bundles
Fig. 1. Experimental set-up for synthesizing B- and N-doped C SWNT bundles.
Time / temperature parameters during the syntheses are shown in Table 1. The experimental runs described in the Table may be divided into four main groups, A, B, C and D, with respect to the yield of doped SWNT bundles. Regimes 1 and 2 form group A, for which mostly untransformed undoped C SWNT bundles were found. The minority of the resultant bundles contained B, although the overall B-content typically did not exceed 2–4 at%. Normally, the N-doping was not observed (the N-content was probably below the detection limit of the EELS spectrometer). Group B, i.e. Regime 3, was found to be optimal for the B- and N-doped SWNT bundle synthesis. In this case nearly 100% of the C starting SWNT bundles were doped with up to |10 at% B and up to |2 at% N. Fig. 2a and b show representative HRTEM images of the B-and N-doped SWNT bundles synthesized under this regime. Typically, resultant SWNTs were stacked in the bundles, although isolated SWNTs were accidentally observed. Numerous EELS spectra taken from the bundles revealed the B / C ratio of #0.1 and the N / C ratio of #0.02 (Fig. 2c). The diameter of the bundles ranged widely. Thin bundles consisting of just a few individual nanotubes (Fig. 2b) or
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
2019
Table 1 Synthesis time and temperature conditions of the experimental runs Regime No.
Temperature, K
Time, min
Group
Comments
1 2
1503 1523
240 30
A A
Mostly undoped C SWNTs, limited number of B-doped SWNTs, B / C#0.04
3
1553
30
B
Primarily B-and N-doped SWNTs, B / C#0.1; N / C#0.02
4 5 6
1533 1573 1623
240 30 30
C C C
Primarily B- and N-doped C SWNTs partially transformed to MWNTs, nanorods and nanoparticles
7 8 9
1703 1723 1773
30 30 30
D D D
Mostly MWNTs, nanorods and nanoparticles of B–C–N and BN, almost all SWNTs are consumed
hundreds of them (Fig. 2a) were seen. The average bundle diameter was found to be |15–20 nm. With increasing synthesis time and / or temperature (Group C) the majority of bundles changed their morphology by collapsing from single-shelled nanotubular network to B–C–N MWNT bundles or nanorods, while the B and N contents in the remaining bundles increased as compared to those produced under Regime 3. However, the B / C and B / N ratios were found to vary irregularly depending on the position of the electron beam along or across the bundles. Accidentally, a few thin bundles consisting of pure BN SWNTs were detected. For the last group (Group D) no SWNT bundles remained, whereas nanoparticles, nanorods and nanoplates of B–C–N and BN materials were found to exist after the synthesis.
3.2. B- and N-doped SWNT formation mechanism During the synthesis the B 2 O 3 powder first melted at ¯723 K and then vaporized with increasing temperature. Since the temperature of the synthesis did not exceed 1773 K we expect that direct vaporization of B 2 O 3 is preferred rather than the formation of dimeric B monooxide vapor (B 2 O 2 ). The B 2 O 3 vapor and N 2 gas flew towards the C SWNT starting bundles and reacted with them. The following chemical reaction may be proposed for the formation of B- and N-doped nanotubes from C nanotubes as the templates [15–17]: xB 2 O 3 1 yN 2 1 (2 1 x 2 2y)C (nanotubes) → 2B x C 12x 2y N y (nanotubes) 1 3xCO
(1)
The above reaction is assumed to proceed while the B 2 O 3 vapor and N 2 gas come in contact with the C SWNT surface. It is obvious that this contact was readily available for the tubes at the bundle’s periphery. These tubes reacted first. The reaction (1) likely leads to the introduction of
topological defects, curling and breaking of the SWNT shells. Two or more damaged neighboring SWNT shells on the bundle periphery may coagulate to form a new shell of the larger diameter due to elimination of dangling bonds. The same mechanism may be considered for the oxidation of the initially thin C SWNT bundles consisting of just a few SWNTs. Therefore, generally, the reaction may lead to the increase of the SWNT diameter, as demonstrated in Fig. 3a and b. Frequently, the diameter of the outer doped SWNTs in the bundles or isolated doped SWNTs increased to 2.1–3.2 nm, as compared to the starting C SWNTs uniform diameter of |1.4 nm. It is emphasized here that increase in the SWNT diameter is a characteristic but not an essential feature of doped SWNTs. B-doped SWNTs exhibiting diameters of |1.4 nm and packed in the bundles were observed early [17] and in the course of this work. Previously, the present authors have assumed that the opening of a nanotube may take place during thermo– chemical treatment due to oxidation of the C atoms by the B 2 O 3 vapor [17]. This assumption is in line with the conclusions drawn earlier by Tsang et al. [18] and Ajayan et al. [19] who studied nanotube oxidation at T .873 K by carbon dioxide and oxygen, respectively. The opening may occur in the SWNT caps containing pentagonal rings of high chemical reactivity or within graphene-like shells at the positions of localized pentagonal or heptagonal defects. This may allow C, B and N atoms to flow along the open tunnels inside the SWNTs or between neighboring SWNTs in a bundle, and across the bundle. The inter-tube tunnels were previously assumed to accumulate intercalating atoms of K [20,21], I [22], and H [23] in C SWNT bundles. However, intercalation of B and N atoms into C SWNT bundles has not been reported to the best of our knowledge. Moreover, a pronounced p*-peak of the B–K edge in the EELS spectrum of doped bundles (Fig. 2c, left-hand peak of the B–K edge) leads us to conclude that B atoms are in the same sp 2 -hybridised state as their C counterparts
2020
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 2. Representative HRTEM images of the B- and N-doped SWNT bundles synthesized at 1553 K over 30 min: (a) A thick bundle; (b) a thin bundle; (c) The representative EELS spectrum taken from a doped SWNT bundle using an electron probe focused down to 1.6 nm in diameter. The positions of B and C and N K-edges are displayed. The magnified result of the background subtraction revealing the pure N K-edge is shown in the inset of (c). The B / C and N / C ratios of |0.1, and |0.02, respectively, were calculated.
(i.e. built-in within the SWNT shells) rather than form atomic chains inside the inter- and / or intra-tube tunnels. The low signal to noise ratio makes resolution of the p* and s* bands difficult for the N K-edge in Fig. 2c. Anyway, we may assume here that the B and N atoms substitute for C atoms in the nanotubular shells as they do for B-doped [16] and BN [15,17] MWNTs produced by the same technique. The distribution of B and N atoms in the present doped SWNT bundles is an unresolved issue. Complications arise
due to the inhomogeneous doping, i.e. difference in chemical compositions between outer and inner SWNTs in the resultant bundles. The measured B / C and N / C ratios in different regions of the bundles produced under optimal Regime 3 (Table 1) were never found to be equal (the B / C ratio was typically much higher than the N / C one). Thus, a simple assumption that we are dealing with sandwich-like structures consisting of pure BN SWNTs on the bundle periphery and undoped C SWNTs in the bundle core is not correct. It is likely that numerous ternary B–C–N SWNTs
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 3. HRTEM images showing increase in diameter for the Band N-doped SWNTs synthesized at 1533 K over 240 min. The SWNT diameters are pointed out with the arrows. (a) An isolated SWNT exhibiting diameter |2.1 nm; (b) an outer tube in the bundle displaying the diameter of |3.2 nm. The starting C SWNT diameters were |1.4 nm.
are formed in the bundle during the syntheses. The trend to form BN-rich SWNTs is small compared to the yield of B-doped C SWNTs with marginal N-doping (Fig. 2c), these results are similar to those obtained earlier for C MWNT doping [4]. Recently it has been observed that B-doping into graphite stabilizes N incorporation and formation of B–N bonds [24]. Formation of solid solutions between graphite and hexagonal BN was also discussed [25]. It is noted that both random and ordered solid solutions of BN and C 2 in an individual SWNT are possible due to the isoelectronic and isostructural properties of BN and C 2 units [4]. B-rich islands (BC 3 stoichiometry) in B-doped tubes and BN nanodomains in composite B–C–N tubes were calculated to be preferential with respect to random distribution of doping atoms [26], although experimental verification of such distribution in the case of SWNT bundles is difficult.
3.3. Packing of doped SWNTs in the bundles It is well documented that SWNTs typically crystallize in bundles or ‘ropes’ due to attraction to each other from van der Waals forces [27]. Fig. 4a and b show the crosssection views of a starting C SWNT bundle and a
2021
Fig. 4. Representative cross-sectional HRTEM images of a starting pure C SWNT bundle grown on a catalytic Ni particle (a) and a B-doped C SWNT bundle synthesized at 1503 K over 240 min in a flowing nitrogen atmosphere (b). The individual SWNTs of interest discussed in the text are pointed out by the arrows and enlarged in the insets. A schematic representation of the honeycomb-like 2D packing of the individual SWNTs in the bundle in (b) is shown in the inset.
marginally B-doped SWNT bundle (|2–4 at% B) respectively. The cross-sectional images of each tube appear as dark contrast closed fringes when the bundles happen to orient in such a way that the tubule axes are approximately parallel to the incident electron beam, i.e. the SWNT shells are tangent to the electron beam ensuring higher C or C and B atom densities and maximum scattering. Thus, these fringes can be directly interpreted as the mean perimeters of the SWNTs at the optimum defocus. Clearly, both images in Fig. 4a and b reveal that SWNT cross-sections in the bundles are not strictly circular. While the starting C SWNTs are packed in a nearly 2D trigonal lattice with an average lattice parameter of 1.7 nm [28] but display irregularly deformed cross-sections, which may be only roughly approximated as circles with a mean diameter of |1.4 nm (i.e. a SWNT in Fig. 4a pointed out by an arrow and enlarged in the inset), some of the individual SWNTs in the resultant doped bundle clearly show hexagonal cross-sections (i.e. a SWNT in Fig. 4b indicated by an arrow and enlarged in the inset). A honeycomb array of
2022
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
the individual tubes in local areas of the bundle with a relatively uniform van der Waals spacing of approximately 0.3 nm is visible. The geometry of such packing implies a shrinking of the distance between neighboring SWNT axes to |1.5–1.6 nm. The effects of cross-section polygonization have been early addressed by Tersoff and Ruoff [29] who predicted that polygonization of the individual SWNT cross-sections in a bundle might be favorable as a result of optimization of the van der Waals inter-tube interactions. These interactions were assumed to press together neighboring SWNTs in a bundle [27]. The calculations are in fact in line with our HRTEM observations in the sense that SWNTs in a bundle core typically looked more polygonized, being bounded on sides by other tubes, but more round in shape on the surface of the bundle. Since the area of a hexagon is less than that of a circle of equal perimeter, the effect of C SWNT polygonizaion and packing into a honeycomb lattice has very recently been used to explain anomalous volume reduction of SWNT material under pressures of up to 20 kbar [30]. However, we stress here that the theoretical threshold diameters of SWNTs in a bundle necessary to ensure energetic stability for noticeable polygonization so far have been considered much higher than those observed by us experimentally in the thermo–chemically treated B-doped SWNT bundles (Fig. 4b). A SWNT diameter exceeding |3 nm was calculated to be necessary for flattening to be significant enough to be visible [29], while SWNTs in Fig. 4b have characteristic cross-section dimensions of approximately 1.2–1.4 nm. To date, the interactions between SWNTs of |1.4 nm in diameter have been considered not sufficiently strong to distort their cylindrical shape [28]. The fact that the experimentally observed polygonization is so strong in the B-doped SWNT bundles agrees with our previous data on faceting of single- and nested fullerenes in the presence of traces of B in their structures [31]. In fact, the honeycomb array of SWNTs in a bundle should optimize the van der Waals interactions between SWNTs by adjusting the uniform inter-tube spacing of |0.3 nm. The particular role of B dopant and / or thermal treatment for the energetics of SWNT bundle 2D packing is an open question yet to be clarified by ab initio calculations.
3.4. Shell number doubling and encapsulation in doped nanotubes The cross-sectional view of the resultant doped bundles frequently shows high contrast fringes inside a SWNT after the thermo–chemical treatment, as shown in Fig. 5. It could be suspected that these fringes are an artifact caused by the specific optical effects of a field emission HRTEM. In order to make it clear we performed the computer simulations for the SWNT HRTEM images, as shown in the inset of Fig. 5. Both highly symmetrical atomic arrangements of the graphitic shells, i.e. armchair (one of
Fig. 5. Cross-sectional HRTEM image of a B-doped SWNT bundle synthesized at 1503 K over 240 min and exhibited inner shells of |0.7 nm in diameter within the outer shells of |1.4 nm in diameter. Note the different contrast for the inner shells in various nanotubes of the bundle. Computer-simulated cross-sectional HRTEM images of the armchair (a) and zigzag (b) SWNTs at the defocus value of 250 nm are shown in the inset for comparison with the experimental HRTEM images.
the C–C bonds is perpendicular to the SWNT axis) and zigzag (one of the C–C bonds is parallel to the SWNT axis), were taken into account (Fig. 5). Obviously, the experimentally observed core contrast in Fig. 5 is much stronger than that caused by the optical effects. In addition, it is noted that the core fringes have a different contrast from one doped nanotube to another in Fig. 5 in spite of the fact that all individual tubes are at the same focusing conditions. Thus it can be concluded that the observed core contrast is not an artifact caused by the optical effects, but it has a physical origin due to the naturally existing atomic densities. Moreover, it may be considered that the contrast intensity is proportional to the number of atoms contributing to the image. Therefore, the various contrast of the inner shells in Fig. 5 may imply a different object length in the direction parallel to the incident electron beam. A plane view of a thermo–chemically treated B-doped thin bundle consisting of just two individual tubes confirmed the statements made above (Fig. 6). The right-hand tube (|1.4 nm in diameter) of the bundle does not show any contrast in its core. By contrast, the left-hand tube (|1.4 nm in diameter) displays an inner shell of just |0.7 nm in diameter which most likely formed during the thermo–chemical treatment, since such contrast has not been observed for any individual SWNT in the starting material (more than seventy HRTEM images were taken from various C SWNT bundles in the starting product). The inner tube has a diameter |0.7 nm, i.e. the theoretical diameter of a C 60 type single fullerene molecule. Accidentally, fullerene-like encapsulates were also observed in the resultant individual doped nanotubes, as shown in Fig. 7a–c. Although the size of these encapsulates is extremely small to allow chemical analysis from
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 6. Plane HRTEM image of a two-tube B-doped bundle synthesized at 1503 K over 240 min. The left-hand tube exhibits an inner shell (d|0.7 nm) formed inside the outer shell (d|1.4 nm).
them, C-based (spheroidal cages in Fig. 7a and b) or BN-clusters (octahedral cage in Fig. 7c, [17]) can be presumed to form inside the B- and N-doped SWNTs after the synthesis. It is known that under thermal treatment the SWNT shell morphology might be modified. Nikolaev et al. [32] have shown that diameter doubling may take place in
2023
heated C SWNTs due to the collapsing of shells from accommodation of defective sites caused by gas phase species attacks at 1673 K. Bonnamy et al. [33] have observed continuous increase in C SWNT diameter with increase in thermal treatment temperature. Recently, Burteaux et al. [34] have studied encapsulation of C 60 molecules into an individual SWNT produced by pulse laser vaporization. These molecules have been found to move freely and even coagulate with each other, leading to the formation of elongated encapsulated fullerenes, i.e. short nanotubes. The mechanism underlying SWNT morphology modification under thermal treatments is a questionable issue. Louchev and Sato [35] have shown that surface diffusion is important for the C MWNT growth originating from a SWNT. Oxidation of present SWNT bundles by the B 2 O 3 vapor may lead to open channel formation in a bundle, as we discussed above. The C and B or N atoms supplied by the surrounding medium may be adsorbed into a SWNT onedimensional channel (intra-tube channel) or migrate to the interstitial channels between the individual SWNT in a bundle (inter-tube channels). Further growth of the internal shell may be fed by surface diffusion along the inner SWNT surface. Based on this mechanism, a penetration depth of the adatoms of the order of 20d–30d (where d is the starting nanotube diameter, equal to |1.4 nm), i.e. |30–40 nm, along the open SWNT channel may be a realistic estimate [36]. The formation of a seed atom ring or a cluster from which the inner encapsulate grows is a statistical nucleation process. This may explain why some tubes do grow encapsulates, i.e. fullerene-like clusters or tubular shells, but some do not (Figs. 5–7). The statistical character of this process is responsible for the length and shape differences of the encapsulates, as revealed by HRTEM contrast difference for various internal objects in the thermo–chemically treated doped SWNT bundles (Figs. 5–7).
3.5. Helicity distribution in the B- and N-doped SWNT bundles
Fig. 7. HRTEM images of individual B- and N-doped SWNTs synthesized at 1553 K over 30 min and displaying fullerene-like encapsulates. (a) and (b) presumably C 60 -based spheroidal encapsulates; (c) presumably a BN-based octahedral encapsulate.
Since a nanotubular object satisfies the weak phase object approximation, its plane image has the maximum contrast where an electron beam meets the most C, B and N atoms. This occurs where the beam is tangent to the SWNT walls. Thus the typical SWNT HRTEM image consists of two parallel dark (0002) fringes whose separation is equal to the mean diameter of the SWNT. However, under favorable optical conditions the atomic structure in the SWNT core region may be resolved, as demonstrated by us in a recent paper [37]. In such cases one may observe characteristic dot-like contrast (if the axis of a SWNT in a thin bundle is happen to orient strictly perpendicular to the electron beam) or the characteristic 0.21 nm fringes of the graphitic lattice, which may be either perpendicular to the tube axis or have an angle |308
2024
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 8. Representative diffraction pattern (DP) taken from a doped SWNT bundle synthesized at 1533 K over 240 min. The regular positions of the diffraction spots for the non-helical armchair (a) and zigzag (z) tubes and the bundle axis are shown.
with respect to it, depending on the shell atomic configuration (zigzag or armchair, respectively) [37]. The traditional point of view that the armchair shells are preferable in SWNT C bundles [38] has recently met strong arguments against it. It was demonstrated by scanning tunneling microscopy (STM) and HRTEM [37,39,40] that there is no preferable orientation in C SWNT bundles and that zigzag, armchair and helical SWNT are common with no particular arrangement dominating. Since the described mechanism of the B- and N-substitution for C does not necessary imply the change in C SWNT shell atomic configuration, it is natural to assume that there should be no helicity selection criteria in the doped SWNTs as well. Indeed, the diffraction pattern (DP)
in Fig. 8 and the HRTEM images in Fig. 9a and b showed that there is no selective helicity. Fig. 8 shows a DP taken from a doped SWNT bundle. Halo-like DP without resolution of the characteristic diffraction spots is seen. The atomic resolution on the individual doped SWNTs confirmed the diffraction data. Fig. 9a and b show atomically-resolved HRTEM images of the individual Band N-doped SWNTs in the local areas of the bundles (the encircled areas in Fig. 9a and b). Computer-simulated SWNT HRTEM images are shown in the insets of Fig. 9 for comparison. The possibility of a slight tilt of the SWNT axis with respect to the figure plane was taken into account in the computer simulations. Horizontal 0.21 nm fringes or those having an angle of |308 with respect to the bundle axis are characteristic of the zigzag or armchair atomic arrangements of the SWNT shells in Fig. 9a and b, respectively. Both zigzag and armchair doped SWNT were atomically-resolved in numerous HRTEM images taken with none of the orientation dominating.
3.6. Limitations of the applied technique for the doped SWNT synthesis With increasing time and / or temperature of the thermo– chemical treatment, as compared to that of the optimal Regime 3, Table 1, the SWNT bundles became unstable and transformed to MWNTs (Fig. 10), polygonal particles and nanorods (Fig. 11a). These nanostructures accumulated much more B and N than the SWNT bundles did. While the B and N contents in doped SWNT bundles produced at high yields were measured to be #10 at% and #2 at%, respectively, the polygonal particles and nanorods in some
Fig. 9. Atomically-resolved HRTEM images of the individual B- and N-doped C SWNTs in the bundles synthesized at 1553 K over 30 min: (a) A zigzag SWNT; (b) an armchair SWNT. The clearly resolved areas are encircled. Corresponding computer-simulated HRTEM SWNT images at the defocus values of 240 nm (a) and 270 nm (b), and structural models are shown in the insets. In the case of the armchair SWNT simulations in (b) the |58 tilt between the figure plane and the SWNT axis was assumed to exist.
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
2025
Fig. 10. HRTEM image demonstrating a B- and N-doped SWNT bundle partially transformed to a B–C–N MWNT during the synthesis at 1533 K over 240 min.
cases displayed pure BN compositions, as shown in Fig. 11b. Negligible traces of oxygen were found in some B- and N-doped SWNT bundles due to the specific features of the synthesis. The levels of the O impurities in the synthesized B-and N-doped bundles were measured in the range of 0–1 at%. Finally, we note that the applied technique may be extended to the production of the C SWNT bundles doped separately by either B or N species by changing nitrogen to argon atmosphere (pure B-doped SWNT bundles) or by excluding B 2 O 3 material during the synthesis (pure Ndoped SWNT bundles).
4. Conclusions Bundles of B- and N-doped C SWNTs were synthesized at high yields during the heating of C SWNTs together with B 2 O 3 in a flowing nitrogen atmosphere. Substitution of B and N for C during SWNT oxidation by the B 2 O 3 vapor is proposed to underlie the formation mechanism. Almost 100% of the starting C SWNT bundles were converted to B- and N-doped SWNT bundles accumulating up to |10 at% B and up to |2 at% N during the synthesis at 1553 K over 30 min. HRTEM analysis of the doped C SWNT bundles revealed the following: 1. The individual SWNTs in a bundle tend to stack in a honeycomb lattice with the average inter-tube spacing of |0.3 nm. 2. Frequently, the diameters of the isolated SWNTs and those in the periphery of the bundles increased to |2.1–3.2 nm, as compared to that of the uniform starting C SWNT diameter of |1.4 nm. 3. Encapsulating of C- or BN-based fullerene-like clusters or growing internal nanotubular shells |0.7 nm in diameter were characteristics of some doped C SWNTs.
Fig. 11. (a) A polygonal particle and nanorods transformed from C SWNT bundles under the synthesis at 1703 K over 30 min. (b) The representative EELS spectrum taken from these nanostructures using an electron beam focused down to 1.6 nm in diameter. The B and N–K edges are shown. The B / N ratio was calculated to be |1.0.
4. No selective helicity was found in the doped SWNT bundles. Both zigzag and armchair SWNTs were found with no arrangement dominating.
Acknowledgements This work was carried out under a project of the Japan Science and Technology Corporation. The authors are grateful to Doctors S. Iijima (Nanotubulites Project, JSTICORP) and F. Kokai (Institute of Research and Innovation) for supplying a C SWNT specimen, to Doctors O. Louchev and A. Matveev (National Institute for Research
2026
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
in Inorganic Materials) for stimulating discussions, and to Doctor J. Hester (Australian Nuclear Science and Technology Organization) for reading the manuscript and providing useful suggestions.
References [1] Iijima S. Helical microtubules of graphitic carbon. Nature 1991;354:56–8. [2] Ajayan P, Iijima S. Smallest carbon nanotube. Nature 1992;358:23. [3] Stephan O, Ajayan PM, Colliex C, Redlich Ph, Lambert JM, Bernier P, Lefin P. Doping graphitic and carbon nanotube structures with boron and nitrogen. Science 1994;266:1683– 5. ¨ [4] Redlich P, Loeffler J, Ajayan PM, Bill J, Aldinger F, Ruhle M. B–C–N nanotubes and boron doping of carbon nanotubes. Chem Phys Lett 1996;260:465. [5] Weng-Sieh Z, Cherrey K, Chopra NG, Blase X, Miyamoto Y, Rubio A, Cohen ML, Louie SG, Zettl A, Gronsky R. Synthesis of BxCyNz nanotubules. Phys Rev 1995;51:11229–32. [6] Zhang Y, Gu H, Suenaga K, Iijima S. Heterogeneous growth of B–C–N nanotubes by laser ablation. Chem Phys Lett 1997;279:264–9. [7] Terrones M, Benito AM, Manteca-Diego C, Hsu WK, Osman OI, Hare JP, Reid DG, Terrones H, Cheetham AK, Prassides K, Kroto HW, Walton DRM. Pyrolytically-grown BxCyNz nanomaterials: nanofibres and nanotubes. Chem Phys Lett 1996;257:576–82. [8] Terrones M, Redlich Ph, Crobert N, Trazobares S, Hsu W-K, Terrones H, Zhu Y-Q, Hare JP, Reeves CL, Cheetham AK, ¨ Ruhle M, Kroto HW, Walton DRM. Carbon nitride nanocomposites: formation of aligned CxNy nanofibres. Adv Mater 1999;11:655–8. [9] Suenaga K, Colliex C, Demoncy N, Loiseau A, Pascard H, Willaime F. Synthesis of nanoparticles and nanotubes with well-separated layers of boron nitride and carbon. Science 1997;278:653–5. [10] Carroll DL, Redlich Ph, Blase X, Charlier J-C, Curran S, ¨ Ajayan PM, Roth S, Ruhle M. Effects of nanodomain formation on the electronic structure of doped carbon nanotubes. Phys Rev Lett 1998;81:2332–5. ¨ [11] Wei B, Spolenak R, Kohler-Redlich P, Ruhle M, Arzt E. Electrical transport in pure and boron-doped carbon nanotubes. Appl Phys Lett 1999;74:3149–51. [12] Hernandez E, Goze C, Bernier P, Rubio A. Elastic properties of C and BxCyNz composite nanotubes. Phys Rev Lett 1998;80:4502–5. [13] Hassanien A, Tokumoto M, Kumazawa Y, Kataura H, Maniwa Y, Suzuki S, Achida Y. Atomic structure and electronic properties of single-wall carbon nanotubes probed by scanning tunneling microscope at room temperature. Appl Phys Lett 1998;73:3839–41. [14] Radovic L, Karra M, Skokova K, Thrower P. The role of substitutional carbon in carbon oxidation. Carbon 1998;36:1841–5. [15] Han W, Bando Y, Kurashima K, Sato T. Synthesis of boron nitride nanotubes from carbon nanotubes by a substitution reaction. Appl Phys Lett 1998;73:3085–7.
[16] Han W, Bando Y, Kurashima K, Sato T. Boron-doped carbon nanotubes prepared through a substitution reaction. Chem Phys Lett 1999;299:368–73. [17] Golberg D, Bando Y, Han W, Kurashima K, Sato T. Singlewalled B-doped carbon, B / N-doped carbon and BN nanotubes synthesized from single-walled carbon nanotubes through a substitution reaction. Chem Phys Lett 1999;308:337–42. [18] Tsang SC, Harris PJF, Green MLH. Thinning and opening of carbon nanotubes by oxidation using carbon dioxide. Nature 1993;362:520–2. [19] Ajayan PM, Ebbesen TW, Ichihashi T, Iijima S, Tanigaki K, Hiura H. Opening carbon nanotubes with oxygen and implications for filling. Nature 1993;362:522–5. [20] Pichler T, Sing M, Knupfer M, Golden MS, Fink J. Potassium intercalated bundles of single-wall carbon nanotubes: electronic structure and optical properties. Sol St Comm 1999;109:721–6. [21] Gao G, Cagin T, Goddard III WA. Position of K atoms in doped single-walled carbon nanotube crystals. Phys Rev Lett 1998;80:5556–9. [22] Fan X, Dickey EC, Pennycock SJ, Grigorian L, Eklund PC, Buczko R, Pantelides ST. Direct observation of Iodine atomic chains in I-doped carbon nanotubes. In: Bailey GW, editor, Microscopy and Microanalysis, Vol. 5, Springer, 1999, p. 144. [23] Ye Y, Ahn CC, Witham C, Fultz B, Liu J, Rinzler AG, Colbert D, Smith KA, Smalley RE. Hydrogen adsorption and cohesive energy of single-walled carbon nanotubes. Appl Phys Lett 1999;74:2307–9. [24] Konno H, Nakahashi T, Inagaki M, Sogabe T. Nitrogen incorporation into boron-doped graphite and formation of B–N bonding. Carbon 1999;37:471–5. [25] Kawaguchi M. B / C / N materials based on the graphite network. Adv Mat 1997;9:615–25. [26] Blase X, Charlier J-C, De Vita A, Car R. Structural and electronic properties of composite BxCyNz nanotubes and heterojunctions. Appl Phys 1999;A68:293–300. [27] Ruoff RS, Tersoff J, Lorents DC, Subramoney S, Chan B. Radial deformation of carbon nanotubes by van der Waals forces. Nature 1993;364:514–6. [28] Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J, Xu C, Lee YH, Kim SG, Rinzler AG, Colbert DT, Scuseria GE, Tomanek D, Fischer JE, Smalley RE. Crystalline ropes of metallic carbon nanotubes. Science 1996;273:483–7. [29] Tersoff J, Ruoff RS. Structural properties of carbon nanotube crystal. Phys Rev Lett 1994;73:676–9. [30] Chesnokov SA, Nalimova VA, Rinzler AG, Smalley RE, Fisher JE. Mechanical energy storage in carbon nanotube springs. Phys Rev Lett 1999;82:343–6. [31] Golberg D, Bando Y, Kurashima K, Sasaki T. Fullerene and onion formation under electron irradiation of boron-doped graphite. Carbon 1999;37:293–9. [32] Nikolaev P, Thess A, Rinzler AG, Colbert DT, Smalley RE. Diameter doubling of single-wall nanotubes. Chem Phys Lett 1997;266:422–7. [33] Bonnamy S, Metenier K, Journet C, Bernier P, de la Chapelle L, Lefrant S, Beguin F. In: Inagaki M, editor, Science and Technology for New Carbons, Nihon Printing Co. Ltd, 1998, p. 122. [34] Burteaux B, Claye A, Smith BW, Monthioux M, Luzzi DE,
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027 Fisher JE. Abundance of encapsulated C60 in single-wall carbon nanotubes. Chem Phys Lett 1999;310:21–4. [35] Louchev O, Sato Y. Nanotube self-organization: formation by step-flow growth. Appl Phys Lett 1999;74:194–6. [36] Louchev O. Personal communication. [37] Golberg D, Bando Y, Bourgeois, Kurashima K. Atomic resolution of single-walled carbon nanotubes using a field emission high-resolution transmission electron microscope. Carbon 1999;37:1858–1860.
2027
[38] Bernaerts D, Zettl A, Chopra NG, Thess A, Smalley RE. Electron diffraction study of single-wall carbon nanotubes. Solid State Com 1998;105:145–9. ¨ JWG, Venema LC, Rinzler AG, Smalley RE, Dekker [39] Wildoer C. Electronic structure of atomically resolved carbon nanotubes. Nature 1998;391:59–61. [40] Henrard L, Loiseau A, Journet C, Bernier P. What is the chirality of singlewall nanotubes produced by arc-discharge? An electron diffraction study. Synth Met 1999;103:2533–6.
Carbon 38 (2000) 2017–2027
Large-scale synthesis and HRTEM analysis of single-walled B- and N-doped carbon nanotube bundles D. Golberg*, Y. Bando, L. Bourgeois, K. Kurashima, T. Sato National Institute for Research in Inorganic Materials, Namiki 1 -1, Tsukuba, Ibaraki 305 -0044, Japan Received 17 November 1999; accepted 21 February 2000
Abstract Bundles of B- and N-doped single-walled carbon nanotubes (SWNTs) containing up to |10 at% B and up to |2 at% N were synthesized at high yields under thermo–chemical treatment of pure C SWNT bundles and B 2 O 3 in a flowing nitrogen atmosphere. The bundles were characterized by means of high-resolution transmission electron microscopy and electron energy loss spectroscopy. The effects of synthesis temperature (1503–1773 K) and time (30–240 min) on the B and N contents and yield of the SWNT bundles were determined. The maximum yield of the B- and N-doped SWNT bundles was obtained under synthesis at 1553 K over 30 min. Atomic structure and morphology of individual SWNTs in the bundles, in particular, packing of doped SWNTs, helicity distribution, encapsulation of fullerene-like clusters, diameter and shell number variations were studied. The synthesized SWNTs in the bundles were stacked in a honeycomb array with the uniform inter-tube spacing of |0.3 nm. No preferable orientation for the graphene-like tubular shells was found, i.e. both zigzag and armchair edges were observed with approximately equal proportions. Frequently, diameter increase took place for the outer tubes in a bundle and for isolated SWNTs. C-based or BN-based fullerene-like encapsulates were observed in individual SWNTs. Carbon oxidation by the B 2 O 3 vapor and B and N substitution for C is thought to underlie the doping of C SWNTs. The substitution reaction temperature–time limits with respect to the morphological stability of B- and N-doped SWNT bundles are finally elucidated. 2000 Elsevier Science Ltd. All rights reserved. Keywords: A. Doped carbons; Carbon nanotubes; C. Transmission electron microscopy (TEM); electron energy loss spectroscopy (EELS); D. Microstructure
1. Introduction Nanotubes made of carbon [1] are expected to bring significant breakthroughs in the technology of electronic and engineering materials of the next millennium. For applications, single-walled nanotubes (SWNTs) are preferred over other nanotubular structures [2], since the number of structural defects in them, which can dramatically affect the properties, is significantly reduced. Doping of C nanotubes with B and / or N [3–8] or preparing sandwich-like structures made of C and BN layers [9] may allow to tailor nanotube electronic [10,11] and mechanical properties [12]. For instance, B-doped nanotubes were found to exclusively exhibit metallic conductivity [10] in contrast to undoped nanotubes whose properties vary between metallic and semiconducting depending on helici*Corresponding author. Fax: 181-298-51-6280. E-mail address: [email protected] (D. Golberg).
ty and diameter [13]. N-doped nanofibres were also expected to be metallic [8]. In addition, B-doping was found to improve graphitization of multi-walled carbon nanotubes (MWNTs) [4]. Also, it is known that B enhances the oxidation resistance of graphite and conventional carbon fibers [14] which is important for applications. Doping of C MWNTs and preparation of composite B–C–N MWNTs are usually achieved by arc-discharge [4] or laser ablation [6] by changing the chemical composition of starting electrodes or targets, respectively. Both processes occur in the highly non-equilibrium conditions which negatively influence the chemical and structural homogeneity of the product. An alternative way of effecting C MWNT chemical composition variations has recently been discovered: A general chemical substitution reaction during which C atoms in the nanotubular network are substituted with B or B and N atoms under nanotube oxidation by the B 2 O 3 vapor in an argon or nitrogen atmosphere [15,16]. Later on, this method has been
0008-6223 / 00 / $ – see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S0008-6223( 00 )00058-0
2018
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
successfully applied for changing chemical composition of C SWNT bundles by the present authors [17]. However, the limited yield and insufficient chemical homogeneity of the doped SWNT product are serious drawbacks which so far have made a detailed characterization of B- and Ndoped SWNT bundles difficult [17]. The present paper reports on the pioneering of large-scale synthesis of B- and N-doped C SWNT bundles by optimization of the synthesis parameters during the substitution reaction and detailed analysis of the resultant doped SWNT product by means of high-resolution transmission electron microscopy paired with electron energy loss spectroscopy.
2. Experimental Bundles of pure C SWNTs produced by Nd:YAG laser ablation either by CarboLex (USA) or JST-ICORP (Japan) were heated together with B 2 O 3 in a flowing nitrogen atmosphere at 1503–1773 K over 30–240 min. The heating was carried out in a vertical induction furnace with a susceptor made of graphite [15–17]. The B 2 O 3 powder was placed in an open sintered graphite crucible and then covered with C SWNTs. The experimental set-up is shown in Fig. 1. The N 2 gas was introduced into the chamber at the ambient pressure from the upper and lower inlets (Fig.
1) and its flows were maintained at 3 l / min (upper flow) and 0.2 l / min (lower flow) during the synthesis. The synthesis temperature was monitored using an optical pyrometer with the accuracy of 610 K. Upon completing the synthesis the product of the reaction was extracted from the crucible, milled in an agate mortar and mixed with CCl 4 . A few drops of the resultant mixture were dripped onto a standard [3 mm carboncoated-cooper grid. High-resolution transmission electron microscopy (HRTEM) was carried out by means of a field emission electron microscope JEM-3000F (JEOL) operating at 300 kV. The microscope had the following characteristics: Spherical aberration coefficient — 1.1 mm; chromatic aberration coefficient — 1.8 mm, point resolution — 0.16 nm; stability of high voltage — 4310 26 ; and tilt angle 6308. HRTEM images were taken at magnification 4310 5 on a negative film at the vicinity of the optimal defocus value of the microscope, — 56 nm. Parallel detection electron energy loss spectroscopy (Gatan 666) was performed with the aim of measuring B and N contents in the product. Normally, an electron beam focused down to [0.5–1.6 nm was placed at different points along and across the SWNT bundles for measuring chemical content variations. Computer simulated SWNT HRTEM images were calculated and displayed using the ‘MacTempas’ software.
3. Results and discussion
3.1. Effect of synthesis parameters on the B /N contents and yield of doped SWNT bundles
Fig. 1. Experimental set-up for synthesizing B- and N-doped C SWNT bundles.
Time / temperature parameters during the syntheses are shown in Table 1. The experimental runs described in the Table may be divided into four main groups, A, B, C and D, with respect to the yield of doped SWNT bundles. Regimes 1 and 2 form group A, for which mostly untransformed undoped C SWNT bundles were found. The minority of the resultant bundles contained B, although the overall B-content typically did not exceed 2–4 at%. Normally, the N-doping was not observed (the N-content was probably below the detection limit of the EELS spectrometer). Group B, i.e. Regime 3, was found to be optimal for the B- and N-doped SWNT bundle synthesis. In this case nearly 100% of the C starting SWNT bundles were doped with up to |10 at% B and up to |2 at% N. Fig. 2a and b show representative HRTEM images of the B-and N-doped SWNT bundles synthesized under this regime. Typically, resultant SWNTs were stacked in the bundles, although isolated SWNTs were accidentally observed. Numerous EELS spectra taken from the bundles revealed the B / C ratio of #0.1 and the N / C ratio of #0.02 (Fig. 2c). The diameter of the bundles ranged widely. Thin bundles consisting of just a few individual nanotubes (Fig. 2b) or
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
2019
Table 1 Synthesis time and temperature conditions of the experimental runs Regime No.
Temperature, K
Time, min
Group
Comments
1 2
1503 1523
240 30
A A
Mostly undoped C SWNTs, limited number of B-doped SWNTs, B / C#0.04
3
1553
30
B
Primarily B-and N-doped SWNTs, B / C#0.1; N / C#0.02
4 5 6
1533 1573 1623
240 30 30
C C C
Primarily B- and N-doped C SWNTs partially transformed to MWNTs, nanorods and nanoparticles
7 8 9
1703 1723 1773
30 30 30
D D D
Mostly MWNTs, nanorods and nanoparticles of B–C–N and BN, almost all SWNTs are consumed
hundreds of them (Fig. 2a) were seen. The average bundle diameter was found to be |15–20 nm. With increasing synthesis time and / or temperature (Group C) the majority of bundles changed their morphology by collapsing from single-shelled nanotubular network to B–C–N MWNT bundles or nanorods, while the B and N contents in the remaining bundles increased as compared to those produced under Regime 3. However, the B / C and B / N ratios were found to vary irregularly depending on the position of the electron beam along or across the bundles. Accidentally, a few thin bundles consisting of pure BN SWNTs were detected. For the last group (Group D) no SWNT bundles remained, whereas nanoparticles, nanorods and nanoplates of B–C–N and BN materials were found to exist after the synthesis.
3.2. B- and N-doped SWNT formation mechanism During the synthesis the B 2 O 3 powder first melted at ¯723 K and then vaporized with increasing temperature. Since the temperature of the synthesis did not exceed 1773 K we expect that direct vaporization of B 2 O 3 is preferred rather than the formation of dimeric B monooxide vapor (B 2 O 2 ). The B 2 O 3 vapor and N 2 gas flew towards the C SWNT starting bundles and reacted with them. The following chemical reaction may be proposed for the formation of B- and N-doped nanotubes from C nanotubes as the templates [15–17]: xB 2 O 3 1 yN 2 1 (2 1 x 2 2y)C (nanotubes) → 2B x C 12x 2y N y (nanotubes) 1 3xCO
(1)
The above reaction is assumed to proceed while the B 2 O 3 vapor and N 2 gas come in contact with the C SWNT surface. It is obvious that this contact was readily available for the tubes at the bundle’s periphery. These tubes reacted first. The reaction (1) likely leads to the introduction of
topological defects, curling and breaking of the SWNT shells. Two or more damaged neighboring SWNT shells on the bundle periphery may coagulate to form a new shell of the larger diameter due to elimination of dangling bonds. The same mechanism may be considered for the oxidation of the initially thin C SWNT bundles consisting of just a few SWNTs. Therefore, generally, the reaction may lead to the increase of the SWNT diameter, as demonstrated in Fig. 3a and b. Frequently, the diameter of the outer doped SWNTs in the bundles or isolated doped SWNTs increased to 2.1–3.2 nm, as compared to the starting C SWNTs uniform diameter of |1.4 nm. It is emphasized here that increase in the SWNT diameter is a characteristic but not an essential feature of doped SWNTs. B-doped SWNTs exhibiting diameters of |1.4 nm and packed in the bundles were observed early [17] and in the course of this work. Previously, the present authors have assumed that the opening of a nanotube may take place during thermo– chemical treatment due to oxidation of the C atoms by the B 2 O 3 vapor [17]. This assumption is in line with the conclusions drawn earlier by Tsang et al. [18] and Ajayan et al. [19] who studied nanotube oxidation at T .873 K by carbon dioxide and oxygen, respectively. The opening may occur in the SWNT caps containing pentagonal rings of high chemical reactivity or within graphene-like shells at the positions of localized pentagonal or heptagonal defects. This may allow C, B and N atoms to flow along the open tunnels inside the SWNTs or between neighboring SWNTs in a bundle, and across the bundle. The inter-tube tunnels were previously assumed to accumulate intercalating atoms of K [20,21], I [22], and H [23] in C SWNT bundles. However, intercalation of B and N atoms into C SWNT bundles has not been reported to the best of our knowledge. Moreover, a pronounced p*-peak of the B–K edge in the EELS spectrum of doped bundles (Fig. 2c, left-hand peak of the B–K edge) leads us to conclude that B atoms are in the same sp 2 -hybridised state as their C counterparts
2020
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 2. Representative HRTEM images of the B- and N-doped SWNT bundles synthesized at 1553 K over 30 min: (a) A thick bundle; (b) a thin bundle; (c) The representative EELS spectrum taken from a doped SWNT bundle using an electron probe focused down to 1.6 nm in diameter. The positions of B and C and N K-edges are displayed. The magnified result of the background subtraction revealing the pure N K-edge is shown in the inset of (c). The B / C and N / C ratios of |0.1, and |0.02, respectively, were calculated.
(i.e. built-in within the SWNT shells) rather than form atomic chains inside the inter- and / or intra-tube tunnels. The low signal to noise ratio makes resolution of the p* and s* bands difficult for the N K-edge in Fig. 2c. Anyway, we may assume here that the B and N atoms substitute for C atoms in the nanotubular shells as they do for B-doped [16] and BN [15,17] MWNTs produced by the same technique. The distribution of B and N atoms in the present doped SWNT bundles is an unresolved issue. Complications arise
due to the inhomogeneous doping, i.e. difference in chemical compositions between outer and inner SWNTs in the resultant bundles. The measured B / C and N / C ratios in different regions of the bundles produced under optimal Regime 3 (Table 1) were never found to be equal (the B / C ratio was typically much higher than the N / C one). Thus, a simple assumption that we are dealing with sandwich-like structures consisting of pure BN SWNTs on the bundle periphery and undoped C SWNTs in the bundle core is not correct. It is likely that numerous ternary B–C–N SWNTs
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 3. HRTEM images showing increase in diameter for the Band N-doped SWNTs synthesized at 1533 K over 240 min. The SWNT diameters are pointed out with the arrows. (a) An isolated SWNT exhibiting diameter |2.1 nm; (b) an outer tube in the bundle displaying the diameter of |3.2 nm. The starting C SWNT diameters were |1.4 nm.
are formed in the bundle during the syntheses. The trend to form BN-rich SWNTs is small compared to the yield of B-doped C SWNTs with marginal N-doping (Fig. 2c), these results are similar to those obtained earlier for C MWNT doping [4]. Recently it has been observed that B-doping into graphite stabilizes N incorporation and formation of B–N bonds [24]. Formation of solid solutions between graphite and hexagonal BN was also discussed [25]. It is noted that both random and ordered solid solutions of BN and C 2 in an individual SWNT are possible due to the isoelectronic and isostructural properties of BN and C 2 units [4]. B-rich islands (BC 3 stoichiometry) in B-doped tubes and BN nanodomains in composite B–C–N tubes were calculated to be preferential with respect to random distribution of doping atoms [26], although experimental verification of such distribution in the case of SWNT bundles is difficult.
3.3. Packing of doped SWNTs in the bundles It is well documented that SWNTs typically crystallize in bundles or ‘ropes’ due to attraction to each other from van der Waals forces [27]. Fig. 4a and b show the crosssection views of a starting C SWNT bundle and a
2021
Fig. 4. Representative cross-sectional HRTEM images of a starting pure C SWNT bundle grown on a catalytic Ni particle (a) and a B-doped C SWNT bundle synthesized at 1503 K over 240 min in a flowing nitrogen atmosphere (b). The individual SWNTs of interest discussed in the text are pointed out by the arrows and enlarged in the insets. A schematic representation of the honeycomb-like 2D packing of the individual SWNTs in the bundle in (b) is shown in the inset.
marginally B-doped SWNT bundle (|2–4 at% B) respectively. The cross-sectional images of each tube appear as dark contrast closed fringes when the bundles happen to orient in such a way that the tubule axes are approximately parallel to the incident electron beam, i.e. the SWNT shells are tangent to the electron beam ensuring higher C or C and B atom densities and maximum scattering. Thus, these fringes can be directly interpreted as the mean perimeters of the SWNTs at the optimum defocus. Clearly, both images in Fig. 4a and b reveal that SWNT cross-sections in the bundles are not strictly circular. While the starting C SWNTs are packed in a nearly 2D trigonal lattice with an average lattice parameter of 1.7 nm [28] but display irregularly deformed cross-sections, which may be only roughly approximated as circles with a mean diameter of |1.4 nm (i.e. a SWNT in Fig. 4a pointed out by an arrow and enlarged in the inset), some of the individual SWNTs in the resultant doped bundle clearly show hexagonal cross-sections (i.e. a SWNT in Fig. 4b indicated by an arrow and enlarged in the inset). A honeycomb array of
2022
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
the individual tubes in local areas of the bundle with a relatively uniform van der Waals spacing of approximately 0.3 nm is visible. The geometry of such packing implies a shrinking of the distance between neighboring SWNT axes to |1.5–1.6 nm. The effects of cross-section polygonization have been early addressed by Tersoff and Ruoff [29] who predicted that polygonization of the individual SWNT cross-sections in a bundle might be favorable as a result of optimization of the van der Waals inter-tube interactions. These interactions were assumed to press together neighboring SWNTs in a bundle [27]. The calculations are in fact in line with our HRTEM observations in the sense that SWNTs in a bundle core typically looked more polygonized, being bounded on sides by other tubes, but more round in shape on the surface of the bundle. Since the area of a hexagon is less than that of a circle of equal perimeter, the effect of C SWNT polygonizaion and packing into a honeycomb lattice has very recently been used to explain anomalous volume reduction of SWNT material under pressures of up to 20 kbar [30]. However, we stress here that the theoretical threshold diameters of SWNTs in a bundle necessary to ensure energetic stability for noticeable polygonization so far have been considered much higher than those observed by us experimentally in the thermo–chemically treated B-doped SWNT bundles (Fig. 4b). A SWNT diameter exceeding |3 nm was calculated to be necessary for flattening to be significant enough to be visible [29], while SWNTs in Fig. 4b have characteristic cross-section dimensions of approximately 1.2–1.4 nm. To date, the interactions between SWNTs of |1.4 nm in diameter have been considered not sufficiently strong to distort their cylindrical shape [28]. The fact that the experimentally observed polygonization is so strong in the B-doped SWNT bundles agrees with our previous data on faceting of single- and nested fullerenes in the presence of traces of B in their structures [31]. In fact, the honeycomb array of SWNTs in a bundle should optimize the van der Waals interactions between SWNTs by adjusting the uniform inter-tube spacing of |0.3 nm. The particular role of B dopant and / or thermal treatment for the energetics of SWNT bundle 2D packing is an open question yet to be clarified by ab initio calculations.
3.4. Shell number doubling and encapsulation in doped nanotubes The cross-sectional view of the resultant doped bundles frequently shows high contrast fringes inside a SWNT after the thermo–chemical treatment, as shown in Fig. 5. It could be suspected that these fringes are an artifact caused by the specific optical effects of a field emission HRTEM. In order to make it clear we performed the computer simulations for the SWNT HRTEM images, as shown in the inset of Fig. 5. Both highly symmetrical atomic arrangements of the graphitic shells, i.e. armchair (one of
Fig. 5. Cross-sectional HRTEM image of a B-doped SWNT bundle synthesized at 1503 K over 240 min and exhibited inner shells of |0.7 nm in diameter within the outer shells of |1.4 nm in diameter. Note the different contrast for the inner shells in various nanotubes of the bundle. Computer-simulated cross-sectional HRTEM images of the armchair (a) and zigzag (b) SWNTs at the defocus value of 250 nm are shown in the inset for comparison with the experimental HRTEM images.
the C–C bonds is perpendicular to the SWNT axis) and zigzag (one of the C–C bonds is parallel to the SWNT axis), were taken into account (Fig. 5). Obviously, the experimentally observed core contrast in Fig. 5 is much stronger than that caused by the optical effects. In addition, it is noted that the core fringes have a different contrast from one doped nanotube to another in Fig. 5 in spite of the fact that all individual tubes are at the same focusing conditions. Thus it can be concluded that the observed core contrast is not an artifact caused by the optical effects, but it has a physical origin due to the naturally existing atomic densities. Moreover, it may be considered that the contrast intensity is proportional to the number of atoms contributing to the image. Therefore, the various contrast of the inner shells in Fig. 5 may imply a different object length in the direction parallel to the incident electron beam. A plane view of a thermo–chemically treated B-doped thin bundle consisting of just two individual tubes confirmed the statements made above (Fig. 6). The right-hand tube (|1.4 nm in diameter) of the bundle does not show any contrast in its core. By contrast, the left-hand tube (|1.4 nm in diameter) displays an inner shell of just |0.7 nm in diameter which most likely formed during the thermo–chemical treatment, since such contrast has not been observed for any individual SWNT in the starting material (more than seventy HRTEM images were taken from various C SWNT bundles in the starting product). The inner tube has a diameter |0.7 nm, i.e. the theoretical diameter of a C 60 type single fullerene molecule. Accidentally, fullerene-like encapsulates were also observed in the resultant individual doped nanotubes, as shown in Fig. 7a–c. Although the size of these encapsulates is extremely small to allow chemical analysis from
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 6. Plane HRTEM image of a two-tube B-doped bundle synthesized at 1503 K over 240 min. The left-hand tube exhibits an inner shell (d|0.7 nm) formed inside the outer shell (d|1.4 nm).
them, C-based (spheroidal cages in Fig. 7a and b) or BN-clusters (octahedral cage in Fig. 7c, [17]) can be presumed to form inside the B- and N-doped SWNTs after the synthesis. It is known that under thermal treatment the SWNT shell morphology might be modified. Nikolaev et al. [32] have shown that diameter doubling may take place in
2023
heated C SWNTs due to the collapsing of shells from accommodation of defective sites caused by gas phase species attacks at 1673 K. Bonnamy et al. [33] have observed continuous increase in C SWNT diameter with increase in thermal treatment temperature. Recently, Burteaux et al. [34] have studied encapsulation of C 60 molecules into an individual SWNT produced by pulse laser vaporization. These molecules have been found to move freely and even coagulate with each other, leading to the formation of elongated encapsulated fullerenes, i.e. short nanotubes. The mechanism underlying SWNT morphology modification under thermal treatments is a questionable issue. Louchev and Sato [35] have shown that surface diffusion is important for the C MWNT growth originating from a SWNT. Oxidation of present SWNT bundles by the B 2 O 3 vapor may lead to open channel formation in a bundle, as we discussed above. The C and B or N atoms supplied by the surrounding medium may be adsorbed into a SWNT onedimensional channel (intra-tube channel) or migrate to the interstitial channels between the individual SWNT in a bundle (inter-tube channels). Further growth of the internal shell may be fed by surface diffusion along the inner SWNT surface. Based on this mechanism, a penetration depth of the adatoms of the order of 20d–30d (where d is the starting nanotube diameter, equal to |1.4 nm), i.e. |30–40 nm, along the open SWNT channel may be a realistic estimate [36]. The formation of a seed atom ring or a cluster from which the inner encapsulate grows is a statistical nucleation process. This may explain why some tubes do grow encapsulates, i.e. fullerene-like clusters or tubular shells, but some do not (Figs. 5–7). The statistical character of this process is responsible for the length and shape differences of the encapsulates, as revealed by HRTEM contrast difference for various internal objects in the thermo–chemically treated doped SWNT bundles (Figs. 5–7).
3.5. Helicity distribution in the B- and N-doped SWNT bundles
Fig. 7. HRTEM images of individual B- and N-doped SWNTs synthesized at 1553 K over 30 min and displaying fullerene-like encapsulates. (a) and (b) presumably C 60 -based spheroidal encapsulates; (c) presumably a BN-based octahedral encapsulate.
Since a nanotubular object satisfies the weak phase object approximation, its plane image has the maximum contrast where an electron beam meets the most C, B and N atoms. This occurs where the beam is tangent to the SWNT walls. Thus the typical SWNT HRTEM image consists of two parallel dark (0002) fringes whose separation is equal to the mean diameter of the SWNT. However, under favorable optical conditions the atomic structure in the SWNT core region may be resolved, as demonstrated by us in a recent paper [37]. In such cases one may observe characteristic dot-like contrast (if the axis of a SWNT in a thin bundle is happen to orient strictly perpendicular to the electron beam) or the characteristic 0.21 nm fringes of the graphitic lattice, which may be either perpendicular to the tube axis or have an angle |308
2024
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
Fig. 8. Representative diffraction pattern (DP) taken from a doped SWNT bundle synthesized at 1533 K over 240 min. The regular positions of the diffraction spots for the non-helical armchair (a) and zigzag (z) tubes and the bundle axis are shown.
with respect to it, depending on the shell atomic configuration (zigzag or armchair, respectively) [37]. The traditional point of view that the armchair shells are preferable in SWNT C bundles [38] has recently met strong arguments against it. It was demonstrated by scanning tunneling microscopy (STM) and HRTEM [37,39,40] that there is no preferable orientation in C SWNT bundles and that zigzag, armchair and helical SWNT are common with no particular arrangement dominating. Since the described mechanism of the B- and N-substitution for C does not necessary imply the change in C SWNT shell atomic configuration, it is natural to assume that there should be no helicity selection criteria in the doped SWNTs as well. Indeed, the diffraction pattern (DP)
in Fig. 8 and the HRTEM images in Fig. 9a and b showed that there is no selective helicity. Fig. 8 shows a DP taken from a doped SWNT bundle. Halo-like DP without resolution of the characteristic diffraction spots is seen. The atomic resolution on the individual doped SWNTs confirmed the diffraction data. Fig. 9a and b show atomically-resolved HRTEM images of the individual Band N-doped SWNTs in the local areas of the bundles (the encircled areas in Fig. 9a and b). Computer-simulated SWNT HRTEM images are shown in the insets of Fig. 9 for comparison. The possibility of a slight tilt of the SWNT axis with respect to the figure plane was taken into account in the computer simulations. Horizontal 0.21 nm fringes or those having an angle of |308 with respect to the bundle axis are characteristic of the zigzag or armchair atomic arrangements of the SWNT shells in Fig. 9a and b, respectively. Both zigzag and armchair doped SWNT were atomically-resolved in numerous HRTEM images taken with none of the orientation dominating.
3.6. Limitations of the applied technique for the doped SWNT synthesis With increasing time and / or temperature of the thermo– chemical treatment, as compared to that of the optimal Regime 3, Table 1, the SWNT bundles became unstable and transformed to MWNTs (Fig. 10), polygonal particles and nanorods (Fig. 11a). These nanostructures accumulated much more B and N than the SWNT bundles did. While the B and N contents in doped SWNT bundles produced at high yields were measured to be #10 at% and #2 at%, respectively, the polygonal particles and nanorods in some
Fig. 9. Atomically-resolved HRTEM images of the individual B- and N-doped C SWNTs in the bundles synthesized at 1553 K over 30 min: (a) A zigzag SWNT; (b) an armchair SWNT. The clearly resolved areas are encircled. Corresponding computer-simulated HRTEM SWNT images at the defocus values of 240 nm (a) and 270 nm (b), and structural models are shown in the insets. In the case of the armchair SWNT simulations in (b) the |58 tilt between the figure plane and the SWNT axis was assumed to exist.
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
2025
Fig. 10. HRTEM image demonstrating a B- and N-doped SWNT bundle partially transformed to a B–C–N MWNT during the synthesis at 1533 K over 240 min.
cases displayed pure BN compositions, as shown in Fig. 11b. Negligible traces of oxygen were found in some B- and N-doped SWNT bundles due to the specific features of the synthesis. The levels of the O impurities in the synthesized B-and N-doped bundles were measured in the range of 0–1 at%. Finally, we note that the applied technique may be extended to the production of the C SWNT bundles doped separately by either B or N species by changing nitrogen to argon atmosphere (pure B-doped SWNT bundles) or by excluding B 2 O 3 material during the synthesis (pure Ndoped SWNT bundles).
4. Conclusions Bundles of B- and N-doped C SWNTs were synthesized at high yields during the heating of C SWNTs together with B 2 O 3 in a flowing nitrogen atmosphere. Substitution of B and N for C during SWNT oxidation by the B 2 O 3 vapor is proposed to underlie the formation mechanism. Almost 100% of the starting C SWNT bundles were converted to B- and N-doped SWNT bundles accumulating up to |10 at% B and up to |2 at% N during the synthesis at 1553 K over 30 min. HRTEM analysis of the doped C SWNT bundles revealed the following: 1. The individual SWNTs in a bundle tend to stack in a honeycomb lattice with the average inter-tube spacing of |0.3 nm. 2. Frequently, the diameters of the isolated SWNTs and those in the periphery of the bundles increased to |2.1–3.2 nm, as compared to that of the uniform starting C SWNT diameter of |1.4 nm. 3. Encapsulating of C- or BN-based fullerene-like clusters or growing internal nanotubular shells |0.7 nm in diameter were characteristics of some doped C SWNTs.
Fig. 11. (a) A polygonal particle and nanorods transformed from C SWNT bundles under the synthesis at 1703 K over 30 min. (b) The representative EELS spectrum taken from these nanostructures using an electron beam focused down to 1.6 nm in diameter. The B and N–K edges are shown. The B / N ratio was calculated to be |1.0.
4. No selective helicity was found in the doped SWNT bundles. Both zigzag and armchair SWNTs were found with no arrangement dominating.
Acknowledgements This work was carried out under a project of the Japan Science and Technology Corporation. The authors are grateful to Doctors S. Iijima (Nanotubulites Project, JSTICORP) and F. Kokai (Institute of Research and Innovation) for supplying a C SWNT specimen, to Doctors O. Louchev and A. Matveev (National Institute for Research
2026
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027
in Inorganic Materials) for stimulating discussions, and to Doctor J. Hester (Australian Nuclear Science and Technology Organization) for reading the manuscript and providing useful suggestions.
References [1] Iijima S. Helical microtubules of graphitic carbon. Nature 1991;354:56–8. [2] Ajayan P, Iijima S. Smallest carbon nanotube. Nature 1992;358:23. [3] Stephan O, Ajayan PM, Colliex C, Redlich Ph, Lambert JM, Bernier P, Lefin P. Doping graphitic and carbon nanotube structures with boron and nitrogen. Science 1994;266:1683– 5. ¨ [4] Redlich P, Loeffler J, Ajayan PM, Bill J, Aldinger F, Ruhle M. B–C–N nanotubes and boron doping of carbon nanotubes. Chem Phys Lett 1996;260:465. [5] Weng-Sieh Z, Cherrey K, Chopra NG, Blase X, Miyamoto Y, Rubio A, Cohen ML, Louie SG, Zettl A, Gronsky R. Synthesis of BxCyNz nanotubules. Phys Rev 1995;51:11229–32. [6] Zhang Y, Gu H, Suenaga K, Iijima S. Heterogeneous growth of B–C–N nanotubes by laser ablation. Chem Phys Lett 1997;279:264–9. [7] Terrones M, Benito AM, Manteca-Diego C, Hsu WK, Osman OI, Hare JP, Reid DG, Terrones H, Cheetham AK, Prassides K, Kroto HW, Walton DRM. Pyrolytically-grown BxCyNz nanomaterials: nanofibres and nanotubes. Chem Phys Lett 1996;257:576–82. [8] Terrones M, Redlich Ph, Crobert N, Trazobares S, Hsu W-K, Terrones H, Zhu Y-Q, Hare JP, Reeves CL, Cheetham AK, ¨ Ruhle M, Kroto HW, Walton DRM. Carbon nitride nanocomposites: formation of aligned CxNy nanofibres. Adv Mater 1999;11:655–8. [9] Suenaga K, Colliex C, Demoncy N, Loiseau A, Pascard H, Willaime F. Synthesis of nanoparticles and nanotubes with well-separated layers of boron nitride and carbon. Science 1997;278:653–5. [10] Carroll DL, Redlich Ph, Blase X, Charlier J-C, Curran S, ¨ Ajayan PM, Roth S, Ruhle M. Effects of nanodomain formation on the electronic structure of doped carbon nanotubes. Phys Rev Lett 1998;81:2332–5. ¨ [11] Wei B, Spolenak R, Kohler-Redlich P, Ruhle M, Arzt E. Electrical transport in pure and boron-doped carbon nanotubes. Appl Phys Lett 1999;74:3149–51. [12] Hernandez E, Goze C, Bernier P, Rubio A. Elastic properties of C and BxCyNz composite nanotubes. Phys Rev Lett 1998;80:4502–5. [13] Hassanien A, Tokumoto M, Kumazawa Y, Kataura H, Maniwa Y, Suzuki S, Achida Y. Atomic structure and electronic properties of single-wall carbon nanotubes probed by scanning tunneling microscope at room temperature. Appl Phys Lett 1998;73:3839–41. [14] Radovic L, Karra M, Skokova K, Thrower P. The role of substitutional carbon in carbon oxidation. Carbon 1998;36:1841–5. [15] Han W, Bando Y, Kurashima K, Sato T. Synthesis of boron nitride nanotubes from carbon nanotubes by a substitution reaction. Appl Phys Lett 1998;73:3085–7.
[16] Han W, Bando Y, Kurashima K, Sato T. Boron-doped carbon nanotubes prepared through a substitution reaction. Chem Phys Lett 1999;299:368–73. [17] Golberg D, Bando Y, Han W, Kurashima K, Sato T. Singlewalled B-doped carbon, B / N-doped carbon and BN nanotubes synthesized from single-walled carbon nanotubes through a substitution reaction. Chem Phys Lett 1999;308:337–42. [18] Tsang SC, Harris PJF, Green MLH. Thinning and opening of carbon nanotubes by oxidation using carbon dioxide. Nature 1993;362:520–2. [19] Ajayan PM, Ebbesen TW, Ichihashi T, Iijima S, Tanigaki K, Hiura H. Opening carbon nanotubes with oxygen and implications for filling. Nature 1993;362:522–5. [20] Pichler T, Sing M, Knupfer M, Golden MS, Fink J. Potassium intercalated bundles of single-wall carbon nanotubes: electronic structure and optical properties. Sol St Comm 1999;109:721–6. [21] Gao G, Cagin T, Goddard III WA. Position of K atoms in doped single-walled carbon nanotube crystals. Phys Rev Lett 1998;80:5556–9. [22] Fan X, Dickey EC, Pennycock SJ, Grigorian L, Eklund PC, Buczko R, Pantelides ST. Direct observation of Iodine atomic chains in I-doped carbon nanotubes. In: Bailey GW, editor, Microscopy and Microanalysis, Vol. 5, Springer, 1999, p. 144. [23] Ye Y, Ahn CC, Witham C, Fultz B, Liu J, Rinzler AG, Colbert D, Smith KA, Smalley RE. Hydrogen adsorption and cohesive energy of single-walled carbon nanotubes. Appl Phys Lett 1999;74:2307–9. [24] Konno H, Nakahashi T, Inagaki M, Sogabe T. Nitrogen incorporation into boron-doped graphite and formation of B–N bonding. Carbon 1999;37:471–5. [25] Kawaguchi M. B / C / N materials based on the graphite network. Adv Mat 1997;9:615–25. [26] Blase X, Charlier J-C, De Vita A, Car R. Structural and electronic properties of composite BxCyNz nanotubes and heterojunctions. Appl Phys 1999;A68:293–300. [27] Ruoff RS, Tersoff J, Lorents DC, Subramoney S, Chan B. Radial deformation of carbon nanotubes by van der Waals forces. Nature 1993;364:514–6. [28] Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J, Xu C, Lee YH, Kim SG, Rinzler AG, Colbert DT, Scuseria GE, Tomanek D, Fischer JE, Smalley RE. Crystalline ropes of metallic carbon nanotubes. Science 1996;273:483–7. [29] Tersoff J, Ruoff RS. Structural properties of carbon nanotube crystal. Phys Rev Lett 1994;73:676–9. [30] Chesnokov SA, Nalimova VA, Rinzler AG, Smalley RE, Fisher JE. Mechanical energy storage in carbon nanotube springs. Phys Rev Lett 1999;82:343–6. [31] Golberg D, Bando Y, Kurashima K, Sasaki T. Fullerene and onion formation under electron irradiation of boron-doped graphite. Carbon 1999;37:293–9. [32] Nikolaev P, Thess A, Rinzler AG, Colbert DT, Smalley RE. Diameter doubling of single-wall nanotubes. Chem Phys Lett 1997;266:422–7. [33] Bonnamy S, Metenier K, Journet C, Bernier P, de la Chapelle L, Lefrant S, Beguin F. In: Inagaki M, editor, Science and Technology for New Carbons, Nihon Printing Co. Ltd, 1998, p. 122. [34] Burteaux B, Claye A, Smith BW, Monthioux M, Luzzi DE,
D. Golberg et al. / Carbon 38 (2000) 2017 – 2027 Fisher JE. Abundance of encapsulated C60 in single-wall carbon nanotubes. Chem Phys Lett 1999;310:21–4. [35] Louchev O, Sato Y. Nanotube self-organization: formation by step-flow growth. Appl Phys Lett 1999;74:194–6. [36] Louchev O. Personal communication. [37] Golberg D, Bando Y, Bourgeois, Kurashima K. Atomic resolution of single-walled carbon nanotubes using a field emission high-resolution transmission electron microscope. Carbon 1999;37:1858–1860.
2027
[38] Bernaerts D, Zettl A, Chopra NG, Thess A, Smalley RE. Electron diffraction study of single-wall carbon nanotubes. Solid State Com 1998;105:145–9. ¨ JWG, Venema LC, Rinzler AG, Smalley RE, Dekker [39] Wildoer C. Electronic structure of atomically resolved carbon nanotubes. Nature 1998;391:59–61. [40] Henrard L, Loiseau A, Journet C, Bernier P. What is the chirality of singlewall nanotubes produced by arc-discharge? An electron diffraction study. Synth Met 1999;103:2533–6.