VOLUME 5, NUMBER 8, AUGUST 2005 © Copyright 2005 by the American Chemical Society
Mechanical Properties of Continuously Spun Fibers of Carbon Nanotubes Marcelo Motta, Ya-Li Li, Ian Kinloch, and Alan Windle* Department of Materials Science and Metallurgy, UniVersity of Cambridge, Pembroke Street, Cambridge CB2 3QZ, U.K. Received April 5, 2005; Revised Manuscript Received June 14, 2005
ABSTRACT We report on the mechanical properties of fibers consisting of pure carbon nanotube fibers directly spun from an aerogel formed during synthesis by chemical vapor deposition. The continuous withdrawal of product from the gas phase imparts a high commercial potential to the process, either for the production of particularly strong fibers or for the economic production of bulk quantities of carbon nanotubes. Tensile tests were performed on fibers produced from the dissociation of three different hydrocarbons, namely, ethanol, ethylene glycol, and hexane, with a range of iron (catalyst) concentrations. The conditions were chosen to lie within the range known to enable satisfactory continuous spinning, the iron concentration being varied within this range. Increasing proportions of single wall nanotubes were found as the iron concentration was decreased, conditions which also produced fibers of best strength and stiffness. The maximum tensile strength obtained was 1.46 GPa (equivalent to 0.70 N/tex assuming a density of 2.1 g/cm3). The experiments indicate that significant improvements in the mechanical properties can be accomplished by optimizing the process conditions.
The unique structure of carbon nanotubes has stimulated great interest in their potential mechanical properties.1-9 Theoretical models have shown that both multiwalled and single-walled nanotubes (MWNTs and SWNTs) should possess very high axial stiffness, with a Young modulus of the order of 1-2 TPa.8,9 Experimental moduli, determined from isolated nanotubes, have been reported as 950 and 1800 GPa, for MWNTs and SWNTs, respectively, although issues remain as to the effective cross sectional area of individual single-wall tubes.9 These potential properties make carbon nanotubes promising candidates for the development of a new generation of high-performance fibers. On the other hand, the fibers currently available are quite diverse in terms of internal * Corresponding author. E-mail:
[email protected]. 10.1021/nl050634+ CCC: $30.25 Published on Web 07/20/2005
© 2005 American Chemical Society
structure, density, and purity and their mechanical properties reported to date are, in all cases, only a fraction of those obtained for individual nanotubes (throughout this paper a “fiber” is as drawn from the reaction zone, possessing a diameter of some tens of micrometers with the cross section comprising 105-106 individual nanotubes). Zhu et al.10 have reported a Young modulus of 49-77 GPa and a tensile strength of about 1.0 GPa for short lengths of SWNT fibers, while Ericson et al.11 have spun fibers from SWNTs in the presence of a solvent to give a Young’s modulus of 120 ( 10 GPa and a tensile strength of 116 ( 10 MPa. A different approach is to incorporate and align the nanotubes in a polymer matrix.12-19 Vigolo et al.13 have produced “coagulation-spun” SWNT fibers with a strength just short of 1 GPa. This process was developed by Dalton et al.13 to make poly(vinyl alcohol)-SWNT (60%) composite fibers which
Table 1. Sets of Synthesis Conditions Useda fiber Fe carrier gas flow set (atom %)b rate (mL/min) I II III IV
0.022 0.026 0.016 0.009
500 (H2) 500 (2H2/3Ar) 1000 (2H2/3Ar) 400 (H2)
injection ratec (mL/h)
hydrocarbon type
7.5 7.5 7.5 2.0
ethanol ethylene glycol (A) ethylene glycol (B) hexane
a Furnace temperature is 1180 °C. b The atomic percentage of iron is with respect to all other atoms of added material including the carrier gas. c This injection rate is for a solution of thiophene and ferrocene in the hydrocarbon liquid. The concentration of thiophene was adjusted to give a Fe/S atomic ratio of ∼0.7.
showed a tensile strength of 1.8 GPa. Compared with the promise of individual tubes, these relatively low values, which can be linked to a lack of nanotube perfection, alignment, and order, present a well-defined challenge to materials processing. We have recently developed a method for producing pure carbon nanotube fibers20 which involves direct spinning of continuous fibers from an aerogel of carbon nanotubes formed in the chemical vapor deposition (CVD) reaction zone. The precursor material is typically a liquid hydrocarbon feedstock with added ferrocene, which forms the iron nanoparticles that act as nucleation sites for the growth of nanotubes, and thiophene which is an established rate enhancer for vapor grown carbon fibers.21 The process is continuous as the carbon nanotube fibers are spun directly from the gas phase. Table 1 outlines the synthesis conditions used to prepare the fiber samples for mechanical property measurement. Each of the four sets of conditions is a combination which led to successful direct CNT fiber spinning from the different carbon sources, while keeping the temperature (1180 °C) and Fe/S ratio constant. Fibers prepared using four different catalyst concentrations were tested for tensile strength. Although three different liquid hydrocarbons were used as feedstock, their nature did not appear critical in determining fiber properties as long as the ratio of injection rates and carrier gas rates was adjusted to ensure successful spinning. The iron content of the feedstock appeared however to have a key influence on structure and properties. A common characteristic of the fibers spun directly from the vapor phase is the nonuniformity of cross sections measured along a given fiber.20 Figure 1 shows short lengths cut from a fiber spun from ethanol. The diameters vary from 10 to about 100 µm. However, the average effective fiber diameters calculated from the TEX measurements, assuming a nominal density of 2.1 g/cm3, range from 2 to 19 µm. Furthermore, where a non-uniform fiber fractures at its minimum cross section, the TEX parameter, as used, will always tend to underestimate strength. (TEX is expressed as fiber mass/unit length (g/km) and thus is an average measure of fiber diameter). The benefit of using TEX arises because the fiber cross sections are often not circular and frequently consist of two or more loosely associated strands. The microstructure of the fibers showed that they consisted almost entirely of carbon nanotubes, the type of which 1530
Figure 1. Optical microscope images of different parts of a fiber showing variation in diameter.
showed a trend with the iron catalyst concentration, the proportion of single (and double) wall tubes increasing as the iron concentration is decreased. At the concentrated iron end of the range, we see predominately smooth multiwall tubes, as shown in Figure 2a. At the dilute end however (type IV), single-wall and double-wall nanotubes predominate (parts b and d of Figure 2). Mixtures of nanotube types are observed in the fibers produced from ethylene glycol (types II and III), with multiwall and single-wall nanotube bundles coexisting in the same regions. Figure 2c (transmission electron microscopy (TEM) image of type III) shows a mixture of tube types. In all cases the nanotubes were long, most fields of view such as those illustrated in Figure 2, showing no ends at all. We estimate that the tubes are at least a few tens of micrometers long, giving an aspect ratio of 1000 or more, but accurate estimates are difficult to make. Fibers in solvents dispersed with ultrasound still show tubes in excess of several tens of micrometers long although the treatment has been shown to reduce nanotube length. Engineering stress-strain curves for the four synthesis conditions covering the range of iron concentrations were obtained by dividing the force by TEX and multiplying by 2.1 g/cm3, which is chosen as it is the density of the graphite. Figure 3 shows the stress-strain curves for examples of the strongest fibers obtained for each condition (I-IV). The fibers had up to 5% residual iron. This factor could have been included in arriving at the nominal density in determining stress in GPa from force/TEX. However, this modest correction would lead to an enhancement in the apparent strength (in GPa) and we chose not to make it at this stage. The curves all show a low modulus initial region. That this is not a grip effect was confirmed by comparing with a standard Kevlar fiber, where the stress-strain curve was linear right from the origin. It is assumed that this initial region corresponds to further orientation of the nanotubes in the fibers and is analogous to the performance of an asspun polymer fiber prior to drawing. The results of all tensile tests performed are summarized in Figure 4 as plots of strength against maximum strain for each fiber. The plot, in common with others for experimental fibers, indicates that the fibers which show a high strain to Nano Lett., Vol. 5, No. 8, 2005
Figure 2. (a) Typical SEM micrograph showing the multiwall nanotubes strands that form the internal structure of fibers of type I (ethanol). (b) Scanning electron microscopy micrograph of a fracture surface of a fiber of type IV pointing out the characteristic branching of singlewall nanotube bundles. (The micrograph was obtained from a region close to the fiber fracture surface.) (c) TEM micrograph of a mixture of multiwall and single-wall nanotubes, from fibers of type III. The insert shows a multiwalled nanotube with a cross section of 30 nm, typical in this microstructure. (d) TEM micrograph of a fiber of type IV, predominantly composed of a mix of single- and double-wall nanotubes. Note that the average size of the Fe particles is smaller for such fibers (arrows).
break do not show high strength and vice versa. The strain to break data do not correlate especially with strength, and they are to some extent dominated by the remarkable behavior of two fibers of type II which drew at a fairly low load to strains of 100% and 135%. Such high strains were unusual and were only seen for type II fiber. Figure 5 shows plots of fiber strength and the elastic modulus (as the steepest part of the stress/strain curve) as a function of the atomic concentration (%) of iron in the total feedstock mix and the carrier gas. The plots show both the highest and average values. The trend is clearly apparent and coupled with the results of the microscopy indicates that lower concentrations of iron lead to a greater proportion of single-wall tubes which in turn correlates with increasing Nano Lett., Vol. 5, No. 8, 2005
strength and stiffness. It is known that iron particles in the range 1 and 3 nm are optimal for the growth of SW and DWNTs.22 The fact that a lower concentration of iron in the process leads to smaller iron particles is not in itself surprising if one views the growth rate of the particles as dependent on the concentration of iron atoms in the vapor phase. There is here also the tacit assumption that once an iron particle has absorbed sufficient carbon to start to extrude a nanotube, then its probability of further growth as a result of collision with other particles is reduced. As the observed mechanical properties of the fibers are still far below experimental and theoretical values reported for individual nanotubes, it is possibly more sensible to look to differences in nanotube organization to account for the 1531
Figure 3. Engineering stress-strain curves for examples of the strongest fibers obtained for each synthesis condition (I-IV). Note that the two graphs are plotted to different scales. The fracture strengths are as follows: type I, 0.28 GPa (0.13 N/TEX); type II, 0.14 GPa (0.07 N/TEX); type III, 0.96 GPa (0.46 N/TEX); type IV, 1.46 GPa (0.70 N/TEX).
Figure 4. Plots of stress on fracture (σf) and extension to break axes, summarizing the mechanical performance of all fibers tested.
observed trends in properties. The axial orientation of the nanotubes observed in the fibers20 will be the result of the process of spinning from the furnace; however, the ability of a fiber to orient will depend critically on the nature of the cross link points which give the aerogel its elastic properties. In the case of single-wall tubes these cross links may be assumed to be the self-association into parallel “ropes” seen in parts b and d of Figure 2. These interchain secondary bonds will have the ability to slip under stress
and will enable alignment on stretching. In the case of MWNTs the associations are more likely to be the result of tubes intertwined through kinking during growth or multidirectional growth from a single catalyst particle (see Figure 2a). Both associations will give a network that is not so amenable to alignment during elongation, with the possibility of some entanglements providing points of stress concentration which will initiate premature fracture. Mechanical properties of polymer fibers are closely associated with the orientation of the component elements, in this case nanotubes. The wind-up stage of the process reported here imparts some orientation, as does the tensile test itself as is indicated by the low initial modulus of the stress strain curves. A program of experiments is planned which will follow the orientation through all the stages of processing and the tensile test itself. The results will be reported in due course. The process of direct spinning of CNT fibers is still under development, and the data reported here represent the first mechanical assessment of the resultant fibers. They are obtained from a series of runs, using different hydrocarbon feedstocks, in which the parameters were optimized to achieve spinning. The values of strength and maximum elastic modulus show a clear trend with the concentration of catalyst precursor in the feedstock; the mechanical property values which increase with decreasing iron also correlate with electron microscopic observations on the same
Figure 5. Plot of best and mean strength and elastic modulus for the fibers from each set of synthesis conditions. Elastic modulus values were defined as the steepest part of the engineering stress-strain curves. 1532
Nano Lett., Vol. 5, No. 8, 2005
samples which show increasing proportions of single and double wall nanotubes with decreasing iron. It has also proved possible to dissolve out most (at least 90%) of the iron with 1 N hydrochloric acid without destroying either the integrity of the fiber or its mechanical properties. This study is proceeding and will be reported in a future publication. Acknowledgment. The authors wish to acknowledge Professor Paul Smith and Dr. K. Feldman, from ETH Zurich, Switzerland, for providing extra facilities for the experimental work and for helpful advice with respect to mechanical testing methods. The authors also wish to acknowledge Thomas Swan and Co. and the Cambridge-MIT institute for the financial support. References (1) Vigolo, B.; Poulin, P. Improved structure and properties of singlewall carbon nanotube spun fibers. Appl. Phys. Lett. 2002, 81, 121012. (2) Baughman, R.; Zakhidov, A.; de Heer, W. Carbon Nanotubessthe Route Toward Applications. Science 2002, 297, 787-92. (3) Delmotte, J.; Rubio, A. Mechanical properties of carbon nanotubes: a fiber digest for beginners. Carbon 2002, 40, 1729-34. (4) Qi, H.; et al. Determination of mechanical properties of carbon nanotubes and vertically aligned carbon nanotube forests using nanoindentation. J. Mech. Phys. Solids 2003, 51, 2213-37. (5) Ruoff, R.; Qian, D.; Liu. W. Mechanical properties of carbon nanotubes: theoretical predictions and experimental measurements. Physique 2003, 4, 993-1008. (6) Abe, H.; Shmitzu, T.; Ando, A.; Tokumoto, H. Electric transport and mechanical strength measurements of carbon nanotubes in scanning electron microscope. Physica E 2004, 24, 42-45. (7) Ajayan, O., Banhart, F. Nanotubes: Strong bundles. Nat. Mater. 2004, 3, 135-36.
Nano Lett., Vol. 5, No. 8, 2005
(8) Li; et al. Tensile properties of long aligned double-walled carbon nanotube strands. Carbon 2005, 43, 31-35. (9) Popov, V. Carbon nanotubes: properties and application. Mater. Sci. Eng., R 2004, 43, 61-102. (10) Zhu, H.; et al. Direct Synthesis of Long Single-Walled Carbon Nanotube Strands. Science 2002, 296, 884-86. (11) Ericson, L.; et al. Macroscopic, Neat, Single-Walled Carbon Nanotube Fibers. Science 2004, 305, 1447-50. (12) Vigolo, B.; et al. Macroscopic Fibers and Ribbons of Oriented Carbon Nanotubes. Science 2000, 290, 1331-34. (13) Dalton, A.; et al. Super-tough carbon-nanotube fibres. Nature 2003, 423, 703. (14) Lau, K.; Hui, D. The revolutionary creation of new advanced materialsscarbon nanotube composites. Composites, Part B 2002, 33, 263-277. (15) Allaoui, S.; Bai, H.; Cheng, J.; Bai, J. Mechanical and electrical properties of a MWNT/epoxy composite. Compos. Sci. Technol. 2002, 62, 1993-98. (16) Thostenson, E.; Chou, T. On the elastic properties of carbon nanotubebased composites: modelling and characterization. J. Phys. D 2003, 36, 573-582. (17) Meguid, S.; Sun, Y. On the tensile and shear strength of nanoreinforced composite interfaces. Mater. Des. 2004, 25, 289-96. (18) Meincke, O.; et al. Mechanical properties and electrical conductivity of carbon-nanotube filled polyamide-6 and its blends with acrylonitrile/butadiene/styrene. Polymer 2004, 45, 739-748. (19) Wang, C.; Guo, Z.; Fu, S.; Wu, W.; Zhu, D. Polymers containing fullerene or carbon nanotube structures. Prog. Polym. Sci. 2004, 29, 1079-1141. (20) Li, Y.; Kinloch, I.; Windle, A. H. Direct Spinning of Carbon Nanotube Fibers from Chemical Vapor Deposition Synthesis. Science 2004, 304, 276-77. (21) Tibbetts, G.; Bernado, C.; Gorkiewicz, D.; Alig, R. Role of sulphur in the production of carbon fibres in the vapour phase. Carbon 1994, 32, 4, 569-576. (22) Moisala, A.; Nasibulin, A.; Kauppinen, E. The role of metal nanoparticles in the catalytic production of single-walled carbon nanotubessa review. J. Phys. Condens. Matter 2003, 15, 42, S3011-35.
NL050634+
1533
