Commercial Production of Fullerenes and Carbon Nanotubes

Raouf O. Loutfya, Timothy P. Lowea, Alexander P. Moravskya and S. Katagirib

a

MER Corporation, Tucson Arizona USA b FIC Corporation, New York NY USA

It has been slightly over ten years since the development of a way to produce macroscopic quantities of fullerene, and the related discovery of fullerene nanotubes. As a result, over 1500 worldwide patents have been filed for the production and applications of these new materials. These applications are so wide ranging that they extend across different industries with products from additives to polymers, photoconductors, photo-resists, and bio-active agents to cosmetics. MER Corporation in Tucson, Arizona joined the ranks of fullerene enthusiasts at the beginning of its discovery by immediately licensing the HuffmanKrtschmer patents. While we are widely recognized as a producer of fullerene and nanotubes, MER has also been active in developing applications for fullerenes and nanotubes. The different applications investigated by MER will be reviewed in subsequent chapters. The overriding factor for the success of any of these applications, however, is the price of fullerenes. However, the price cannot come down markedly until large-scale applications are found. To introduce the first large-scale application an organization had to take a leap of faith and initiate the large-scale low-cost production. Mitsubishi/FIC Corporations has been a leader and pioneer in recognizing the need to support large-scale production effort to realize the fullerene and nano-technology commercialization dream. It is now our opportunity to realize the commercial applications. The present status of the scale-up production effort of fullerenes and the different nanotubes will be presented in this chapter.

1. Fullerene Production

The family of fullerenes includes the hollow cage all-carbon molecules having a convex closed-shell structure containing arbitrary numbers of hexagonal and exactly twelve pentagonal faces. They are synthesized abundantly in the carbon arc, hydrocarbon flame or field-induced hot carbon plasma and are originally embedded in the soot product. Fullerenes can be extracted from the soot due to appreciable solubility in some organic solvents. Separation of the various members of the fullerene family is accomplished by chromatography. The most stable fullerenes and have been produced in amounts of up to 23 wt % in the total harvested soot in the arc process [1], although typical technological yields are on the order of 815% at different production facilities. The higherorder fullerenes, of which and are the most abundant, typically make up about 24 wt % of the arc-produced soot. It is interesting to note that the molar ratio remains constant at 5.06 within about 1% accuracy in any carbon arc process in helium gas atmosphere, regardless of run conditions or graphite rod thickness [2]. Soon after the famous discovery of the arc process in 1990, MER Corporation was the first to offer fullerenes commercially. The mixed fullerenes that are extracted from the soot were the first product available, and after the development of chromatographic
E. (ed.,) Perspectives of Fullerene Nanolechnology, 3546. 2002 Kluwer Academic Publishers. Printed in Great Britain.

Raouf O. Loutfy, Timothy P. Lowe, Alexander P. Moravsky, et al.

separations on alumina with hexane eluent, purified and were offered. Subsequent improvement in chromatographic methods increased the separation capacity by a factor of several hundred. The first primitive arc reactors were operated manually, with one graphite rod at a time being loaded and vaporized. Automated computercontrolled reactors (Figure 1), which operate without any attendance for several days, were constructed in 1992, and the output of mixed fullerenes and purified and was increased dramatically. Further increases in fullerene production rate were expected by employing thicker graphite rods in larger, more powerful arc reactors. However, fullerene production in carbon arcs was generally believed to be a non-scalable process, in the sense that the use of the rods of larger diameter leads to a rapid reduction in fullerene yield. Guided by detailed mechanistic considerations of the carbon arc process [1], we have proved experimentally, that this is not the case, by vaporization of 3-inch (~76 mm) diameter graphite rods to routinely produce over 11 wt % fullerenes soot product. The scaled-up reactor (Figure 2) is currently in use for commercial fullerene production in Osaka, Japan. With several applications of fullerenes about to flourish, the scale-up of the arc process has reduced the cost of fullerenes to a level that will stimulate these new uses of fullerenes, and will make fullerene products viable alternatives to existing products that use other materials. A further increase in vaporized rod diameter and in arc power is technically feasible but does not appear to be economically justified. The large-scale arc reactor developed by MER Corporation (Figure 2) has been demonstrated to be close to the optimum size and power for fullerene synthesis in an arc, and will probably remain the largest reactor in the world for a long time. In the arc process, the next logical step for increasing fullerene output will be the automation of large-scale arc reactors, which will render the production process nearly continuous. While the arc process is presently the most efficient means for high fullerene production rates, other methods that have been investigated for fullerene synthesis may soon be
36

Commercial Production of Fullerenes and Carbon Nanotubes

adapted for commercial production. To become competitive to the arc process, these alternative techniques would have to make use of their potential to be operated continuously at a level of hundreds of kilowatts useful power per production unit, provided their undesirable properties such as low inherent yield and specific production rate, high operating cost and/or high capital expenditure are obviated. The field-induced high-intensity plasmas are an attractive alternative to the arc process; inexpensive powder, non-conductive carbon or hydrocarbon gas could be used as starting material, and product collection could be automated and continuous. MER is undertaking technical feasibility studies for this method. Other known techniques, such as the use of low-pressure hydrocarbon flames or laser vaporization of carbon have only a slight chance of becoming competitive on a large scale, mainly because of technical difficulties. The flame technique possesses the potentially important advantage of being an exothermic process, but it has too low a specific production rate to ever reach the fullerene output level available with a scaled-up arc process. The hundreds of kilowatt continuous power lasers are prohibitively expensive to be used for fullerene production. In conclusion, it becomes obvious that fullerenes can be produced by virtually any process capable of generating and appropriate curing of carbon vapor, and that many opportunities are not yet explored. A new highly efficient process may be just around the corner.
2. Synthesis of Carbon Nanotubes (MWNTs, SWNTs and DWNTs) and Vapor-Grown Carbon Fibers (VGCFs) by the Arc and CVD Techniques Arc-Grown MWNTs

2.1.

MWNTs are an all-carbon fullerene structure composed of concentric nested graphene cylinders forming an average misorientation angle of 0 to the tube axis. Arc-grown MWNTs are produced without metal catalysts, and are found in the core material of the
37

Raouf O. Loutfy, Timothy P. Lowe, Alexander P. Moravsky, et al.

deposit that is formed on the cathode during vaporization of the anode. A transmission electron microscopy (TEM) image of arc-grown MWNTs is shown in Figure 3. They typically have outer diameters in the range of 350 nm, with 15 nm as the average. The inner diameter is in the 14 nm range, and is most commonly 2 nm. The aspect ratio is around 100, and both ends are capped. Arc-produced MWNTs are perfectly straight and have few defects. They therefore have high mechanical, thermal and electrical conductivity. The Youngs modulus of MWNT is calculated to be ~10001500 GPa and has been measured at 10003700 GPa. The measured tensile strength is 1163 GPa for the outer layer of the MWNT. The thermal conductivity along the length of the tube is ~1500 W/mK, and the resistivity is 2.1.1. Scaled-up production of arc-grown MWNTs MWNTs can be regarded as a useful by-product of fullerene synthesis in the arc. Their distribution, orientation, quality and content in cathode deposits [3, 4] depend primarily on the radial temperature distribution within the arc gap, which in turn is defined by the operating conditions of the arc reactor. While limited amounts of MWNTs are found in the cathode deposits produced during fullerene production, the optimized conditions for MWNT production give approximately 40 wt % MWNTs in the soft core material of the cathode deposit. TEM examination of representative samples show the major contaminants to be multi-layer polyhedral particles (MPPs) and various kinds of graphitic particles. The cathode deposit core is a soft, black fibrous material that is easily separated from the fused carbon shell, which contains few nanotubes. MER/FIC offers the separated core material on a kilogram scale, and it has proven useful as a conductive additive 38

Commercial Production of Fullerenes and Carbon Nanotubes

for plastics, as electron emitters in electron guns, and as probe tips in atomic-force microscopy, among many other uses under development. The laboratory-scale process has proven to be scalable with the production of arcgrown MWNTs in a large-scale reactor built by MER. While the larger-scale reactors have been found to produce lower yield products than small-scale reactors, MWNTs of more than 95% purity have been obtained in research scale samples (Figure 4) by purification of the cathode deposit core material using a procedure which decomposes the graphitic carbon and MPP contaminants. Coincidentally, the purification process also opens the ends of many of the MWNTs. This procedure has clear potential for scale-up to process arc-grown MWNT, and may foster research into effective ways to fill MWNTs with useful materials.
2.2. Catalytic MWNTs

MWNTs are also produced by catalytic pyrolysis of hydrocarbons over metals (catalytic CVD method). Catalytically grown MWNTs are of about the same outer and inner diameters (370 nm and 210 nm) as arc-grown nanotubes, but are usually much longer CCVD-grown tubes are often capped with a metal nanoparticle at the growing end, and may contain metal inclusions in the inner channel. Catalytic MWNTs abound in structural defects, resulting in bends and kinks between straight segments. Therefore their mechanical strength and conductive properties are lower than those of arc-produced MWNTs (sometimes by several orders of magnitude, depending on the amount of defects). Catalytic MWNTs are preferable to arc-produced MWNTs when high aspect ratio is more important than high structural strength or high conductivity.
39

Raouf O. Loutfy, Timothy P. Lowe, Alexander P. Moravsky, et al.

2.2.1.

Scaled-up production of MWNTs by the catalytic pyrolysis of hydrocarbons

At MER, a scaled-up version of a reported process [5] that employs the pyrolytic decomposition of xylenes over a floating ferrocene-derived catalyst produces nearly pure MWNTs, with small amounts of residual iron catalyst. A straightforward treatment to remove iron catalyst nanoparticles yields a product that is nearly 100% MWNTs. The length of these catalytic tubes can reach 1 mm, which is two orders of magnitude longer than the arc-produced MWNTs, while the tube diameter is about the same in both products. The length and diameter of the mass-produced catalytic MWNTs can be controlled over a very wide range, 100nml mm and respectively, by varying the process parameters, as opposed to arc-produced tubes that have a fairly stable size distribution independent of arc conditions. A true continuous process for catalytic MWNT production has been recently developed at MER Corporation. 2.2.2. Synthesis of large arrays of aligned MWNTs by CVD using a floating catalyst

In a system similar to the one where catalytic MWNTs are synthesized, large area (up to uniform arrays of MWNTs are grown that are perfectly aligned normal to a flat support (Figure 5). It has been demonstrated that conductive supports, like n-doped silicon wafers, can be employed, and it was shown that there is good electrical contact between the MWNTs and the support. The thickness of the layer of aligned MWNTs

40

Commercial Production of Fullerenes and Carbon Nanotubes

can be varied from a few micrometers to one millimeter, with surface density of tubes of approximately 30 billion per Aligned MWNT arrays on conductive supports with the tube ends free of iron catalyst particles have also been produced.
2.2.3. Vapor-grown carbon fibers (VGCFs)

The vapor-grown carbon fibers (VGCFs) are near axially symmetric lamellar structures composed of conical-shaped graphene-sheet-based elements. The average misorientation angle between a graphene plane and the tube axis can be in the range Accordingly, the cone angle may vary between 0 and 180. The latter extreme corresponds to a platelet-type structure, ideally composed of a stack of flat graphene sheets. A herring-bone-type structure comprises either intact cones or truncated cones having flat or convex caps. If the cap is absent, a hollow internal cavity is formed and the structure is considered to be a conical layer nanotube (CLNT). In real herring-bone and CLNT structures, the angle can vary within a layer in the radial direction, matching to the shape of the metal catalyst particle from which it formed, and the layers may comprise various defects and discontinuities. With the cone is transformed into a cylinder and VGCFs are referred to as MWNTs, although these tubes can be much thicker than arc-grown MWNTs (a few micrometers in diameter) due to sidewall growth, and they generally possess less orderly structured outer layers. Multiple exposed edges appear on the outer walls of the VGCFs, and the linear density of edges increases with increasing The many exposed edges make VGCFs good candidates for intercalation of the interlayer spaces with small molecules or ions, so they may find use in gas storage or Li-ion electrochemical cells. The edges can also bind transition metal ions, making VGCFs a promising catalyst support. These are the most promising areas for VGCF applications, since their mechanical and conductive properties are poor compared to those of carbon nanotubes and polymer-based carbon fibers.
2.2.4. RF plasma-assisted CVD synthesis of aligned arrays of VGCFs

Perfectly aligned arrays (Figure 6) of VGCFs with herring-bone internal structure have been deposited normal to the surface of electrically conducting supports by the RF plasma-assisted CCVD method [6], which is similar to the microwave plasma CCVD technique that was first reported in [7]. These practically pure VGCFs are typically up to ten times thicker (20200 nm in diameter) than concentric MWNTs. They are less conductive than MWNTs, but are much better adapted to the adsorption and release of chemicals from between the graphene layers. The growing ends of as-produced VGCFs contain nickel or iron microparticles depending on the catalyst employed, which can be removed by treatment with a mineral acid. The diameter and surface density of fibers produced are controlled by the method of catalyst preparation and by the process parameters. Under certain conditions some MWNTs are produced as a component of aligned arrays whose major composition is herring-bone VGCFs.
41

Raouf O. Loutfy, Timothy P. Lowe, Alexander P. Moravsky, et al.

2.2.5.

Bulk production of VGCFs with abundant random exposed edges

Randomly oriented pure VGCFs of herring-bone and platelet internal structure (Figure 7) are obtained at MER Corporation in bulk quantities (kg) by the catalytic pyrolysis of a mixture of ethylene or carbon monoxide with hydrogen over an ironcopper catalyst. This process is described in detail in [8].
2.3. Single-walled carbon nanotube (SWNT)

SWNTs are a fullerene structure made of a graphene sheet rolled into a seamless cylinder having carbon caps on both ends. The diameter of SWNTs produced by the arc, laser or catalytic CVD process is in the range 15 nm, and can be controlled by experimental conditions. The aspect ratio can be as large as Figure 8 is a TEM micrograph of arcgrown SWNTs. SWNTs made by any of the synthetic methods are essentially defect-free structures that possess outstanding mechanical and conductive properties. The tubes are highly elastic and are always buckled or kinked rather than broken if bent. Theoretical estimates predict 12001700 GPa for the value of the Youngs modulus and 3000 W/m K for the axial thermal conductivity. The metallic type of SWNTs possess a very low resistivity of both by theoretical estimates and direct measurements.
2.3.1. Scaled-up arc synthesis of SWNTs

The laboratory-scale synthesis of SWNTs in the arc using Ni/Co based catalyst has been subjected to an extensive set of mechanistic studies [9], leading to the consistent
42

Commercial Production of Fullerenes and Carbon Nanotubes

production of SWNT product with 40 vol % yield of SWNTs. As in the scale-up of fullerene production by increasing the diameter of the reactant rod, there is a general consensus that yield will decrease as the rod diameter increases. However, research at MER has determined that 25 mm diameter rods packed with the Ni/Co based catalyst can be used to increase production with a minimal decrease in yield. An efficient purification 43

Raouf O. Loutfy, Timothy P. Lowe, Alexander P. Moravsky, et al.

procedure employed at present at MER yields research quantities (~10g per month) of more than 90 wt % pure SWNTs (Figure 9). The procedure enables ~30% recovery of pure tubes even from low-yield SWNT starting materials. Therefore, the scale-up of the purification process will make these materials valuable, provided they are produced in abundance at low cost. The control of SWNT properties has proven to be difficult. Variation of arc conditions and/or catalyst composition has some effect on the tube parameters, but more experimentation is required to define the effects. A sulfur-promoted Fe/Co/Ni-based catalyst was recently found [10] that is capable of producing more than 60 wt % yield of SWNTs with very large diameter (about 4.0 nm on the average), which could have special applications. Further optimization of the synthesis parameters and consequent scaling-up of the process are expected to finally give much more efficient production of these high-yield thick tubes. 2.4. Double-walled carbon nanotubes (DWNTs)

DWNTs are a fullerene structure made of two nested graphene cylinders with both cylinders capped at the ends (see Figure 10). The outer diameter of DWNTs is in the range of 25 nm, while the inner diameter is 14 nm, depending on the synthesis conditions. The inter-wall distance (~0.39 nm) is slightly larger than the distance in MWNTs. The length of DWNTs can reach DWNTs are produced in a carbon arc in a technique that is similar to SWNT production, with modifications to the catalysts and atmosphere employed [10]. The outer and inner walls of a tube are defect-free graphene sheets, and the tube probably has similar high mechanical strength and conductive properties to those found in SWNTs. The tubes are highly elastic and buckle at bending, and since they are stiffer than SWNTs, they are typically found in a less curved and tangled state. The 44

Commercial Production of Fullerenes and Carbon Nanotubes

values for mechanical strength, thermal and electrical conductivity for DWNTs are not yet defined.
2.4.1. Selective synthesis of DWNTs in the arc

Selective synthesis of double-walled carbon nanotubes (DWNT) is efficiently accomplished using a slight variation in the same Fe/Co/Ni catalyst composition and preparation procedure used for SWNTs [10]. High-resolution TEM (HRTEM) analysis of typical samples has shown that the ratio of DWNT/SWNT is higher than 30 and the yield of DWNTs in as-produced fibrous material exceeds 70 wt %. The outer diameter of these DWNTs ranges from 3.2 to 4.9 nm, and the size distribution is sharply peaked at 3.8 nm [10]. The scaling-up of this arc synthesis method is underway to make DWNTs available for extensive research work on anticipated applications in cold field emission, hydrogen storage, lithium batteries and other technologies.
3. Conclusion

Different methods were investigated experimentally and theoretically to produce fullerenes and various types of nanotubes (SWNT, arc-MWNT, VG-MWNT, CVD-aligned MWNT, and arc-DWNT). Scale-up production of many of these products was accomplished. Scale-up schemes for all other materials are underway. Accordingly, the limiting factors for the commercialization of the fullerenes-based materials is now finding the right (killer) applications. In Chaps. 2229 some of the MER/FIC applications development efforts are described.
45

Kaouf O. Loutfy, Timothy P. Lowe, Alexander P. Moravsky, et al.

References
1. A. V. Krestinin and A. P. Moravsky. Mechanism of Fullerene Synthesis in the Arc Reactor, Chem. Phys. Lett. 286 (1998) 479485. 2. A. P. Moravsky, A. G. Ryabenko, P. V. Fursikov and A. A. Ryabenko. An Invariant of Carbon Arc Synthesis of Fullerenes, Fullerene Sci. Technol. 6 (1998) 453. 3. N. A. Kiselev, A. P. Moravsky, A. B. Ormont and D. N. Zakharov. SEM and HREM Study of the Internal Structure of Nanotube Rich Carbon Arc Cathode Deposits, Carbon 37 (1999) 10931103. 4. S. Seraphin, D. Zhou, J. Jiao, J. C. Withers and R. Loutfy. Effect of Processing Conditions on the Morphology and Yield of Carbon Nanotubes, Ibid. 31 (1993) 685689. 5. R. Andrews, D. Jacques, A. M. Rao, F. Derbishire, D. Qian, X. Fan, E. C. Dickey and J. Chen. Continuous Production of Aligned Carbon Nanotubes: A Step Closer to Commercial Realization, Chem. Phys. Lett. 303 (1999) 467474. 6. N. A. Kiselev, A. P. Moravsky, E. V. Dreval, J. L. Hutchison, D. N. Zakharov, J. Sloan, A. Franco, M. Hecht and R. O. Loutfy. The Fine Structure of Aligned Carbon Micro- and Nano-Fibers Synthesized by RF Plasma-Enhanced CVD Technique, Carbon (in press). 7. Z. F. Ren, Z. P. Huang, J. W. Xu, J. H. Wang, P. Bush, M. P. Siegal and P. N. Provencio. Synthesis of Large Arrays of Well-Aligned Carbon Nanotubes on Glass, Science 282 (1998) 11051107. 8. N. M. Rodriguez, A. Chambers and R. T. K. Baker. Catalytic Engineering of Carbon Nanostructures, Langmuir 11 (1995) 38623866. 9. R. O. Loutfy, T. P. Lowe, J. L. Hutchison, N. A. Kiselev, D. N. Zakharov, E. P. Krinichnaya, V. E. Muradyan, B. P. Tarasov and A. P. Moravsky. A Dissolution-Precipitation Model for SingleWalled Carbon Nanotubes Formation in the Arc, IWFAC99 (1999) 109. 10. J. L. Hutchison, N. A. Kiselev, E. P. Krinichnaya, A. V. Krestinin, R. O. Loutfy, A. P. Moravsky, V. E. Muradyan, E. D. Obraztsova, S. V. Terekhov and D. N. Zakharov. Double-Walled Carbon Nanotubes Fabricated by Hydrogen Arc Technique, Carbon 39 (2000) 761770.

46