Carbon Nanotube Fibres : Science and Technology Transfer

Transcription

1 18 Carbon Nanotube Fibres : Science and Technology Transfer PROF. ALAN WINDLE Department of Materials Science and Metallurgy, Charles Babbage Road, Cambridge, U.K. ahw1@cam.ac.uk Abstract A unique process for the spinning of carbon nanotube fibre directly from the CVD reaction zone will be described, as will the structure and properties of the product, covering, in particular, recent developments in the quest for electrical conductivity at a level to replace copper. The process is inherently cheap, and given sufficient scale may well rival conventional carbon fibre in terms of cost, as well as providing an alternative property spectrum. A personalised account of the technology transfer efforts and the issues involved, will lead to a view of possibility for industry and society. Keywords : commercialization, CVD, direct spinning. 1. Introduction One of the starting points of the nano technology revolution was the discovery, in 1985, very much from left field, of carbon bucky balls, or Buckminster Fullerine, to give it its full name. Of the family, C 60 was the most prominent member, and the molecule s picture on the cover of Nature in which it was announced, represented a huge step for nano science and the potential applications which would flow from it. Nobel prizes also followed for Curl, Kroto and Smalley. The most important step in the related carbon nanoform of nanotubes was announced in two papers by Ijima 1,2 in 1991 and Whereas there had been numerous reports of sub micron carbon nano fibrils, it was the Ijima papers which defined the structure of the nanotubes and in the second paper related single wall nanotubes to the already famous buckyballs. Within a few years the science of soot had been promoted from a

2 Cinderella to a Princess, to become a hugely respectable area which attracted leading scientists and technologists. Theorists 3 began to predict properties and establish methods of defining the structural types of nanotubes. In particular, it was shown that particular types of single wall carbon nanotubes, should show metallic type conductivity, and more than this, that they could behave, in an ideal world, as quantum wires, where their resistance would not depend on length! Another signal step was an early measurement 4 of the mechanical properties of individual nanotubes. It showed that the strength of each monolayer graphitic cylinder (which today we would call a graphene layer) approached the theoretic estimate rather well. The excitement which followed was in part fuelled by the fact that to tether a satellite in geostationary orbit would require a cable of this type of strength, some 10x greater than the strongest fibre known. However, this hype was presenting an almost impossible challenge to Materials Science, that is to realize a property in a bulk, and thus useful, material which is as good as that of its nanoscale building blocks. It also failed to address the question as to how one would actually make a tether some 60,000 miles long! While nanotubes have many potential uses, the alignment of them into fibres is a valuable one to consider as the material then has the same symmetry (uniaxial) as its nano-scale building blocks. A number of methods have been developed to make fibres from carbon nanotubes. The pioneering process was probably that of coagulation spinning 5 in which a solution of carbon nanotubes, surfactant stabilized, is injected into a flowing polymer solution, which triggers coagulation as a floppy gel. The resultant fibre is really a composite with the polymer, so in that sense is not purely of CNTs. However, the process has demonstrated promising properties and is still under development with a company base (CANOE) 6. Another approach, which produces fibres without additionally added polymeric material, is the spinning of carpetlikearrays of nanotubes grown from a catalyst deposited on a flat 19

3 20 substrate. It has proved possible to pull these nanotubes off sideways, in a way which ensures they are sufficiently entangled or otherwise attached to their neighbours. The result is a thin, aligned film of nanotubes, which can then be twisted into a fibre. The process is particularly suited to development on a laboratory scale, although routes to scale up for commercial high volume fibre production, still appear elusive. A different approach is based on the observation that suspensions of nanotubes in liquids, at sufficiently high concentrations, form lyotropic liquid crystalline phases, in which the level of inter tube entanglements is minimized. 7 The spinning method developed 8 is somewhat analogous to that for the manufacture of Kevlar, with the fibre being drawn from the liquid crystalline solution, and the solvent, typically chlorosulphonic acid, dissolved out of the fibre which is then dried and spooled. Finally, a method which combines both the synthesis of carbon nanotubes and their spinning into one fibre in a single step without the involvement of liquids has been developed here in Cambridge. It is referred to as direct spinning. This process was first announced in and is under continual development. It is described in greater detail below, and is the subject of the case study in industrial commercialization which follows. 2. Direct spinning of CNT Fibres from the Reaction Zone : The Process The process itself is a so-called floating catalyst method. A catalyst precursor, in this case ferrocene, is continually injected along with the source of carbon, typically a hydrocarbon such as ethanol, toluene or methanol. It will also work with natural gas as the carbon supply. The process is illustrated in Fig. 1. The breakdown of the ferrocene leads to the formation of iron catalyst particles in the gas stream, which will tend to grow by collision. The process operates at 1250 C, which is an unusually high temperature compared with many others used for the synthesis of nanotubes, especially those involving growth from a substrate.

4 One result of operating at this temperature is that no carbon nanotube growth occurs from catalyst deposit on the side walls, which is due to a process known as thermopheresis, where nanoscale particles of catalyst experience a net molecular thrust away from the hot reactor walls when within the mean free path (~1 m) of the surface. Another consequence of operating at such a high synthesis temperature is that the nanotube growth reaction is very fast and very long nanotubes are formed within a matter of seconds, entangling to form a gas-based gel (aerogel). This elastic smoke 9 is then continuously withdrawn from the reactor, condensed by spraying with atomized acetone and then wound continuously as a 10 m diameter fibre (Fig. 2). One needs to note that the reactants are injected into the reactor at around 400 C, but rapidly heat to 1250 C within 10 seconds or so. During this period the iron catalyst particles are formed, and both sulphur and some reactive carbon, sufficient to nucleate nanotubes. A key aspect of the process is the addition of sulphur. Without it, although some nanotubes are formed, they are not formed particularly fast and the nanotubes do not grow to the point 21 (a) (b)

5 22 (c) Fig 1 (a) Diagram of the direct spinning process, the nanotubes are synthesised in the gas phase to form an aerogel. They are then continuously would out of the reactor, to form an aligned aggregate which can be subsequently condensed into a fibre; (b) the aligned aggregate of nanotubes being wound from the reactor; (c) the process being demonstrated. where they can form a mechanically coherent aerogel which can be withdrawn as a fibre. The exact role of sulphur is not fully understood. It is known from the technology of cast irons to act as a surfactant, and enhance edgewise growth of graphitic particles within in the microstructure10. It is also known as a promoter in field of reactions involving carbonaceous products.11,12. Certainly, it is crucial to this process. It also seems that if both carbon and sulphur are available as soon as possible after the initial liberation of the iron from the ferrocene, then it is possible to grow single wall nanotubes, and spin fibres comprising of just this type13.

6 23 Acetone atomizer Fig 2 Spaying atomized acetone on the emerging fibre causes, through surface tension forces, the fibre to densify to ~ 1g/cc and a diameter of ~10 micron 3. Structure and Properties of Direct Spun Fibres The structure of the fibre produced consists of highly aligned carbon nanotubes, whether single, double or multiwall or a mixture. Fig. 3 shows a scanning electron micrograph of a fibre. The aligned entities here are in fact not individual nanotubes, which are far too small to see at this magnification, but bundles of aligned tubes, bundles of perhaps 50 nanotubes. One such bundle is shown in Fig. 4(a) in a high resolution transmission electron microscope (HRTEM), with an individual nanotube in the shown to the right - in this case a largish diameter double wall nanotube, which almost certainly will be collapsed into a dog-bone cross section 14. The perfection of orientation can be defined by X-ray diffraction. Fig. 4(b) shows an X-ray diffraction pattern of a fibre consisting of collapsed and stacked double wall nanotubes. The concentration of the inter graphene layer spacing diffraction peak, 002, onto the equator is very obvious The continuous ring at this scattering angle, represents unoriented carbonaceous material, almost certainly not nanotubes.

7 24 1 m Fig. 3 SEM micrograph showing aligned bundles of carbon nanotubes in a partially condensed CNT fibre. The bundles form a network. The individual nanotubes comprising the bundles are not resolved. (Courtesy of Dr Thurid Gspann) Winding rate = 30 m/min (a) (b) Fig 4 (a) High resolution TEM micrograph of a bundle of carbon nanotubes extracted from the fibre, with a single, double wall nanotube to the right. (b) Synchrotron transmission wide angle X-ray diffraction pattern of the fibre showing the high degree of orientation of the graphitic layers (002 reflection). The continuous rings represent graphitic material present as an impurity.

8 Also apparent in Fig. 3 are a few catalyst residue particles. These particles have not grown nanotubes, as the ones which did grow nanotubes would be an order of magnitude smaller or more. In fact, the vast majority of iron added is not itself responsible for nanotube growth, although a reduction in the added proportion of ferrocene soon leads to unspinnable material. Overall the structure consists of carbon nanotubes rather perfectly mutually aligned with bundles. It appears from a series of painstaking HRTEM observations that the nanotubes themselves are especially long, with a length/diameter ratio in excess of 100,000, 14. The bundles form a continuous network, as sub-sets of nanotube break away from one bundle to join a neighbouring one. Where process conditions are set to make single wall material, the evidence is clear from HRTEM, Fig. 5(a), while the existence of Radial Breathing Modes (RBMs) in Raman spectroscopy provides an added confirmation, Fig. 5(b). The different types of 25 (a) (b) Fig. 5 (a) High resolution TEM micrograph of a bundle of single wall carbon nanotubes extracted from the fibre synthesized under conditions favouring single wall material. (b) Peaks in the low wavenumber region of a Raman spectrum, confirming the presence of single wall nanotubes.

9 26 impurity in direct spun CNTs have been reviewed by Chai and Kumar 15. Residual catalyst particles, both those responsible for nanotube growth and those not, are one type of defect, and when particularly large (>0.1 m) they disrupt the organization of the nanotube bundles, leading to local deviation from near perfect orientation, and compromising, in particular, the axial stiffness of the fibre. These residual iron particles are often seen to be coated with carbonaceous material, which can involve very short, highly defective, multi wall nanotubes. A further type of extraneous defect, observed on some occasions, is an overcoat of polymeric material on the bundles, although not penetrating the bundles themselves. It is surmised that it is formed from a synthesis reaction amongst the reactants, possible as the material is cooling on extraction from the reactor. The axial mechanical properties of the fibre depend critically on the process conditions. Optimization leads to fibres which approach 2N/tex strength and 100N/tex in stiffness. (N/tex is exactly the same as specific stress, with units of GPa/SG where SG is the ratio of the fibre density to that of water, i.e. the density in g/cc). These values are similar to those of Kevlar 49, DuPont s renowned aramid fibre. The key difference in mechanical properties however, is that this CNT fibre is yarn-like as it consists of bundles of nanotubes, with a low shear strength between them. The failure mode in tension is, not surprisingly, the result of the bundles failing in shear giving a classic fibrous fracture surface Fig. 6(a). The big advantage of yarn like properties, is that the fibre is exceptionally tough in terms of its resistance to crack propagation across the fibre, as any crack is immediately blunted as it spreads axially up and down the weak interfaces. The fibre is thus resilient to bending and general handling stress. One performance indicator in this regard is knot efficiency where the strength in tension of a knotted fibre is expressed as a percentage of the strength of the pristine fibre. Fig. 6(b) shows that, in comparison with other high performance fibres, this CNT material

10 is remarkably resistant to bending stresses 16. In particular it is far superior to conventional carbon fibre, which for individual ~ 10 µm filaments, breaks easily if bent at all sharply. CNT fibre can also be fed directly into a loom and woven, as shown in Fig. 6(c). 27 (a) CN (c) (b) Fig 6 (a) Fibrous fracture surface of a 10 m diameter CNT yarn-like fibre. (b) Knot efficiency, the strength of a fibre with a single over hand knot in it expressed as a percentage of the strength without the knot. (c) Cloth woven from carbon nanotube fibre. Each woven strand consists of ~ 500 CNT fibres. The orange region is polyester which was attached to the end of the CNT strands which were of limited length. (Courtesy, Dr Fiona Smail). The fibre also appears to be a ready component for composite material. If the fibres are pre aligned, then the matrix can be infiltrated. However, as the fibres are yarn-like the matrix also penetrates the fibre itself, ensuring an excellent fibre/matrix bond. The axial properties of a series of composites [17] indicate that the reinforcement is at least as good as a law of mixtures would suggest. What is more, while the individual yarn-like fibre has little mechanical resistance in compression, as the individual

11 28 bundle components buckle under load, infiltration and cure of the epoxy system endows the fibre composite with compressive stiffness similar to that seen in tension. A fracture surface from a tensile sample is shown in Fig. 7. The crack has propagated through the cured epoxy matrix leaving a very flat surface, indicative of little energy absorption, however, the fracture of each fibre shows evidence of significant local deformation, possibly associated with the local shear and pulling out of epoxy coated nanotube bundles. Fibre to same scale Fig 7 Tensile fracture surface of a CNT fibre/epoxy composite containing 27% by weight of fibre. There is little evidence for localized deformation associated with the fracture of the matrix (brittle) however, there is evidence of considerable local plastic deformation within the fibres, with bundles of nanotubes coated with infiltrated resin apparent on the surface. Note that the fibre cross sections are not particularly circular [17]. The fibre shows values of electrical conductivity two orders of magnitude worse than copper (per unit area), although it is more like one order of magnitude short if one makes a specific property

12 comparison, dividing by density. In some cases fibre spun under conditions which give single wall nanotubes, can show encouragingly higher values of electrical conductivity, between one third and one half of that of copper for the specific property comparison, when they consist predominately of nanotubes with chiralities which will endow the tubes with metallic rather than semiconducting chiralities.. The objective now is clearly to see to what extent this property can be improved further, as CNT fibre may present an intriguing alternative to copper as an electrical conductor. Theory suggests that perfect, metallic single wall nanotubes, suspended in vacuum, should demonstrate ballistic conductivity. In this case, the resistance of a nanotube is predicted to be independent of its length! Of course, reality is far removed from these ideal conditions, but the challenge to materials science is there. Finally, the fibres appear to show excellent thermal conductivity, with values being realized which are three times better than copper in absolute terms and nearly 25 times better when divided by density. These values may open up the possibility of heat cables which will move heat around complicated structures or machinery, and can be wired rather as one would an electrical circuit. 4. Commercialization Perhaps one of the responsibilities of an applied scientist, a Materials Scientist in this case, is to play a part in the process of technical transfer out of a research environment into an industrial one so that the science may be of benefit to commerce, and thus to society at large. Staying with direct spun CNT fibres as an exemplar, we will follow through some of the challenges being encountered in moving a radically new material into the market place. Firstly, it seems that for a new type of material to gain traction, in the minds of those who might be in a position to provide scale-up funds, there has to be a good argument to say that 29

13 30 not only will it need to exhibit better properties, but it will have to be cheaper as well. Just one or the other is not so impressive. There is also another factor, that of the level of investment in the status quo. For a large company to commit to a radically new material product, it will need to assure itself that the new advance will not prejudice any of its existing businesses. As an example, a completely new plant to make traditional carbon fibre, where it also includes synthesis and spinning of the polymeric precursor, may cost something approaching $0.5 billion. Our laboratory process to make direct-spun carbon nanotube fibre, will, on a good day, make about 20 km of single strand fibre, but that only weighs about 1g. It may be a good basis for optimizing properties, understanding the science of a nano-yarn and establishing basic process parameters, but there is obviously much further to go before quantities which industry would even see as a simple demonstrator can be made available. The origins of the fibre process stem from earlier collaborative work between the Cambridge research group and a small, specialist chemicals company in the North of England called Thomas Swan Ltd. 18. The company, which had licensed Cambridge University owned patents, then built a plant to make and market single wall carbon nanotubes, mainly in powder form, as a speciality chemical which sustains a thriving business in this commercial area. In Cambridge, the step to make fibre directly from the reaction zone, with important encouragement from Thomas Swan, was patented and announced in the journal Science 9. The next stage was to build a vehicle to drive the technology transfer into the fibre industry. A University spin-out company, Q-Flo, was founded, with significant University ownership, in It then, funded by its directors, set about finding an industrial partner to push forward the first stage of scale up. For some time the company entertained a whole series of potential partners to the Laboratories in the Department of Materials Science in Cambridge. Some visitors were obviously just scoping what was going on, while others were very enthusiastic,

14 but were not permitted to proceed by their company boards, who saw the opportunity as being too early stage. 31

15 32 Fig 8 Slides from a presentation made to businesses in Texas in To a surprising large degree the business model projected then (third slide) has worked out in practice. With the science base at Cambridge University in place, a partnership with Plasan now producing income to Q-Flo, and routes to second phase growth with co-producers and distributers now under negotiation. Q-Flo was presented with a great opportunity, where it was offered one of the sought-after pitches at a Rice University (Houston) business interface day. This was achieved on behalf of the Q-Flo through the good offices of the UK Consulate in Houston. Q-Flo s presentation was very well received and afterwards there was a scrum for further information. As a matter of historical record, Fig. 8 comprises three slides from that presentation. However, the huge US enthusiasm began to evaporate when enquirers discovered that Q-Flo was UK based and did not already have a presence in the US, let alone Texas. Q-Flo saw the period which followed as one of kissing frogs, but it was not looking good, directors pockets were nearly empty, and the Company was on the point of considering a fire sale when a viable partner came back with a serious offer. The next step forward came with the founding of a joint venture company,

16 TorTech with the new partner, Plasan - a protective armour supplier 20. Plasan is based in Israel with plants in France and the US. The TorTech scale up facility has now achieved approximately x100 in material output rate compared to the laboratory rig, and is poised, in building further new international partnerships, to move towards achieving the next significant scale up stage. 5. Future Prospects To bring a new product to market, especially if that product is a new material, where cost will be inversely related to the scale of manufacture, is a daunting task. Nevertheless, carbon nanotube fibre represents one of the potential benefits of the so-called nano revolution. On the mechanical side it represents, at least, a radically new form of carbon fibre, which with its yarn-like character, shows a toughness in bending far in excess to that of the traditional material. Its fracture process is not a brittle one, but involves considerable local energy lass, so its performance will be much less defect sensitive. One might speculate that it is the fibre s potential as an electrical and thermal conductor which offers the greatest long term benefit, with the possibility of replacing metals such as copper, with a cheaper and, incidentally, stronger alternative. However, the challenge to materials science here, is a classical one, namely to translate into a useful material form the brilliant properties shown by the individualbuilding blocks, in this case the nanotubes. To achieve electrical conductivity which would challenge copper, not only will the fibre have to be made from single wall nanotubes, a step now routinely achievable in the laboratory, but these nanotubes will have to have chiral structures which make them metallic rather than semiconducting. Of the various possible chiralities only a third are metallic, so it is important to learn how to produce fibre consisting of single wall nanotubes only of the desired type. This objective is still to be achieved, although there are some highly encouraging, if not easily reproducible,reports

17 34 The process is being scaled up to the point where samples can be made available, however the next step will be to achieve investment to enable moves towards production on the tonnage scale. One benefit of the process described here is that it is in principle scalable, not simply by multiplying up the number of reactors, but by building a much larger reactor, possibly of radical geometry, where the cost benefits of scale, such as thermal efficiency and recycling of both carrier gas and any unreacted hydrocarbon, will begin to tell. The process, is intrinsically cheap, and will run using natural gas as the major carbon source. There is huge potential here, if only it can be realized. Acknowledgements This article has drawn freely on the research of author's group over recent years. In most cases the work is acknowledged through the papers here cited. However, special thanks are due to Dr. Thurid Gspann for permission to reproduce here unpublished data which is Fig 4, and to Dr Fiona Smail who kindly made Fig 6(c) available. References 1. Iijima,S. (1991) Nature354, Iijima,S. & Ichihashi, T, (1993) Nature363, Saito, R., Dresselhaus, G., Dresselhaus, M.S. (1998) Physical Properties of Carbon Nanotubes, Imperial College Press, London 4. Yu, M.F., Lourie O, Dyer M.J, Moloni K., Kelly T.F. & Ruoff, R.S. (2000) Science 287, Vigolo, B., Penicaud, A., Coulon, C., Sauder, C., Pailler, R., Jouirnet, C., Bernier, P., & Poulin, P. (2000) Science, 290, Song, W., Kinloch, I. A., & Windle, A.H. (2003) Science, 302, Erikson, L.M., Fan, H., Peng, H., Davis, V.A. & Zhou, WS. (2004) Science,305, Li, Y-L., Kinloch, I.A. & Windle, A.H. (2004)Science, 304,

18 10. Skaland, T., Grong, O., & Grong, T. (1993) Metallurgical Transactions A, 24A, Barreiro,A., Kramberger, C., Rummeli, M.H., Gruneis, A., Grimm,D., Hampel, S., Gemming,T., Buchner, B., Bachtold, A., & Pichler, T. (2007) Carbon45, Katsuki, H., Matsunaga, K., Egashira, M., et al (1981) Carbon19, Sundaram, R.M., Koziol, K.K. & Windle, A.H. (2011) Advanced Materials23, Motta, M., Moisala, A., Kinloch, I.A. & Windle, A.H. (2007) Advanced Materials 19 (20) Chai, H. G., & Kumar, S. (2008) Science, 319, Vilatela, J.J. & Windle, A.H. (2010) Advanced Materials, 22, Mora, R. J. Vilatela, J.J. & Windle, A.H. (2009) Composites Science and Technology, 69,