NANO LETTERS
Catalytically Grown Carbon Nanotubes of Small Diameter Have a High Young’s Modulus
2005 Vol. 5, No. 10 2074-2077
Branimir Lukic´,† Jin Won Seo,† Revathi R. Bacsa,‡ Sandrine Delpeux,§ Franc¸ ois Be´guin,§ Geoffroy Bister,| Antonio Fonseca,| Janos B. Nagy,| Andra´s Kis,† Sylvia Jeney,† Andrzej J. Kulik,† and La´szlo´ Forro´*,† Institute of Physics of Complex Matter, Ecole Polytechnique Fe´ de´ rale de Lausanne, 1015 Lausanne, Switzerland, Centre InteruniVersitaire de Recherche et d’Inge´ nierie des Mate´ riaux, UniVersite´ Paul Sabatier, 31062 Toulouse Cedex 4, France, CRMD, CNRS-UniVersite´ , 1B rue de la Fe´ rollerie, 45071 Orle´ ans Cedex 02, France, and Faculte´ s UniVersitaires Notre-Dame de la Paix, 61 rue de Bruxelles, 5000 Namur, Belgium Received June 3, 2005; Revised Manuscript Received July 21, 2005
ABSTRACT Experimental studies of carbon nanotubes (CNTs) obtained through different synthesis routes show considerable variability in their mechanical properties. The strongest CNTs obtained so far had a high Young’s modulus of 1 TPa but could only be produced in gram scale quantities. The synthesis by catalytic chemical vapor deposition, a method that holds the greatest potential for large-scale production, gives CNTs with a high defect density. This leads to low Young’s modulus values below 100 GPa for multiwall CNTs. Here we performed direct measurements of the mechanical properties of catalytically grown CNTs with only a few walls and find a Young’s modulus of 1 TPa. This high value is confirmed for CNTs grown under two different growth conditions where the synthesis parameters such as the hydrocarbon source, catalyst material, and the synthesis temperature were varied. The results indicate that the observed difference in the Young’s modulus for the catalytically grown CNTs with high and low numbers of walls is probably related to the growth mechanism of CNT.
Due to their exceptional mechanical properties, carbon nanotubes (CNTs),1 either single wall or multiwall, should be ideal reinforcing fibers for composite materials. In particular, it has been shown that their Young’s modulus can have values of ∼1 TPa,2,3 which makes CNTs one of the stiffest materials available. They can sustain very high purely elastic deformations4 up to a breaking strength of several tens of GPa.5 Mainly, they are produced by three methods:6 arc-discharge, laser ablation, and chemical carbon vapor deposition (CCVD). Currently, only the first two methods offer CNTs with the expected mechanical properties,5,7 but their production is limited to only gram quantities. CCVD is an established method in the synthesis of carbon fibers and filaments and is the most promising in terms of large-scale production, due to the relative ease of up-scaling the preparation and purification methods.8 It can produce CNTs in quantities of kilograms to tons at temperatures * Corresponding author. E-mail:
[email protected]. Tel: ++41(21)6934306. Fax: ++41(21)6934470. † Ecole Polytechnique Fe ´ de´rale de Lausanne. ‡ Universite ´ Paul Sabatier. § CRMD, CNRS-Universite ´. | Faculte ´ s Universitaires Notre-Dame de la Paix. 10.1021/nl051034d CCC: $30.25 Published on Web 08/30/2005
© 2005 American Chemical Society
typically around 700 °C in contrast to arc discharge where the synthesis process occurs at ∼3000 °C. However, in terms of structure, CCVD-grown CNTs are generally inferior to the CNTs produced by arc discharge and the incorporation of defects in the CNT structure can even be observed in situ.9 They also show poorer mechanical properties, and multiwall CNTs have a Young’s modulus below 100 GPa.10 Hence the improvement of the CCVD-based production toward CNTs with a low defect density, and thus with a high Young’s modulus, is an important issue to drive the industrial application of CNTs in reinforced composites forward. So far, it is still unclear how defect density relates to the growth of CNTs during the CCVD process. An established model of the growth mechanism describes CNT formation by first dissociating hydrocarbon into carbon on the surface of a metallic catalyst particle. Second, carbon is incorporated up to a saturating concentration into the bulk of the catalyst. Eventually, the precipitation of carbon from the saturated particle leads to the formation of CNTs.11 We have demonstrated in a previous study12 that during the CCVD process the growth conditions play a minor role in influencing the defect density of multiwall CNTs, since their Young’s
modulus stayed below 100 GPa on changing their production parameters (e.g., catalyst material and synthesis temperature). However, the diameter, i.e., the number of walls, of CNTs used in the study was quite big varying between 20 and 50 nm. In general, one would expect that incorporation of defects in the CNT structure during growth is more likely to occur for multiwall CNTs than for single- or double-wall CNTs where dislocations13 or the incomplete closure of the outermost graphene sheets9 can hinder structure formation. A higher number of walls can also host more point defects that might not be removed completely by thermal annealing during growth.14 Hence, CNTs with a low number of walls should have better mechanical properties. This assumption has not yet been confirmed, and in the only study published up to now15 a surprisingly low Young’s modulus of 10 GPa for CCVD-grown single-wall CNTs was reported. Here we report direct measurements of the mechanical properties of CCVD-grown CNTs with a low number of walls (i.e., between one and four). Additionally, we prepared two batches16-18 of CNTs synthesized under two different growth conditions in order to see whether they would influence the mechanical properties of CNTs. A detailed structural characterization was made by transmission electron microscopy (TEM) of CNTs produced from both batches, which we refer to as batch I and batch II.19 For batch I (Figure 1a), the outer diameter of our CNTs was in the range of 1.1-3.5 nm and their inner diameters in the range of 0.65-2 nm. More than 70% of CNTs in batch I had two walls, while the rest had one or three walls.16 For batch II (Figure 1b), the CNTs’ outer diameter was between 2 and 16 nm. About 50% of CNTs were double-wall (outer diameter 1.6-4.0 nm). The other half was composed of CNTs with three walls (ca. 25%; outer diameter 3-12 nm) and four walls (ca. 25%; outer diameter 4-16 nm). As such, our samples consisted predominantly of double-wall CNTs. In the representative TEM images of Figure 1, the wellgraphitized wall structures can be seen with wall numbers varying from two to four. Unlike thick multiwall CNTs, these CNTs preferentially formed bundles in which individual nanotubes are held together by van der Waals forces. A cross section of such a bundle can be seen in the lower right corner of Figure 1a. Direct mechanical measurements were done by atomic force microscopy (AFM) using the method introduced by Salvetat et al.20,21 There, a nanotube bundle lies over a hole of a porous alumina membrane, with its major part still in contact with the membrane surface. Parameters such as the bundle diameter, D, and the distance, L, between the clamped ends of the bundle were directly measured from the AFM images (Figure 2a). Simultaneously, the vertical deflection δ of the CNT was measured versus the force F applied to a point midway along its free-standing length (Figure 2b). The deflection followed the expression δ ) FL3/(192EBI), where I is the second moment of area of the bundle (I ) πD4/64) and EB is the effective bending modulus.22,23 EB is a measure of the mechanical response, and in the case of an individual nanotube, it is equivalent to the Young’s modulus Y that Nano Lett., Vol. 5, No. 10, 2005
Figure 1. TEM images of CNTs from batch I (a) and batch II (b).
describes the elastic elongation of individual CNTs. However, for a CNT bundle, the mechanical response also includes the shear modulus, G, that quantifies the sliding between nanotubes within one bundle. It can be accounted for by the following relation22 1 1 10 D2 ) + EB Y 3G L2
(1)
Equation 1 predicts that the contribution from Y dominates for small bundle diameters, while the contribution of G increases for larger diameters, when 2D/L g (G/Y)1/2. This sliding mechanism has the same origin as the sliding between graphene planes in graphite and has been well characterized 2075
Figure 2. CNTs studied by AFM: (a) AFM image of a CNT bundle adhered to an alumina membrane with a part suspended above the hole (image area ∼ 1 µm2); (b) a typical deflection vs force curve obtained from a series of AFM images, under varying load.
Figure 3. Mechanical measurements of CNTs. Bending modulus versus diameter for CNT bundles for batch I (0) and batch II (b). Theoretical value of Y ) 1 TPa (ref 2) for a single CNT is indicated by a dashed line. Inset. Shear modulus, G (D > 5 nm), extracted from the experimental data by assuming Y ) 1 TPa.
for single-wall CNT bundles grown by arc discharge where a Y and a G of the order of 1 TPa and 1 GPa respectively were found.20 In Figure 3, measurements of EB for batch I (0) and batch II (b) with a CNT bundle diameter, D, ranging from 4 to 30 nm are plotted. As can be seen, EB decreases by more than two orders of magnitude with increasing D. The large uncertainty in EB (∼50%) mainly comes from the uncertainty in determining L due to the irregular pore shape of the supporting membrane.20 The highest value for EB is measured for D < 5 nm and results in a Young’s modulus on the order of 1 TPa,24 in accordance with theoretical predictions.2 The sliding between individual CNTs in a bundle is described by the term including the shear modulus, G, in eq 1. It starts to dominate for D > 5 nm when the value of EB decreases abruptly to below 100 GPa. By taking Y ) 1 TPa to calculate G from eq 1, we obtained values below 1 GPa regardless of the bundle diameter as is indicated in the inset of Figure 3. 2076
This suggests one common sliding mechanism for all CNTs in a bundle, which is independent of the number of CNTs per bundle. The high Young’s modulus obtained for CNT bundles with small diameters of both batches is comparable to values for single CNTs made by laser ablation or by arc discharge,5,7 strongly pointing to a low density of defects in their structure. As previously reported, CCVD-grown multiwall CNTs with more than four walls exhibited a Young’s modulus below 100 GPa.10,12 Hence, we can conclude from the present study that the diameter has a strong influence on the Young’s modulus (and on the defect density). The potential functional dependency between these two quantities could improve our still incomplete understanding of the CNT growth mechanism in the CCVD process. Only few functional relations have been established to date, including the influence of the catalyst particle size on the CNT diameter25 and the inverse scaling of the growth rate with the CNT diameter.26,27 Note that our results suggest that other changes in the growth conditions, like the material of the catalyst or the synthesis temperature, play a less significant role in determining the defect density, this was also shown to apply for multiwall CNTs.12 In conclusion, we have shown that CNTs produced with CCVD can have Young’s moduli up to 1 TPa, which is the highest reported value of a direct measurement of the Young’s modulus. It is in marked contrast with the mechanical properties of CNTs with high numbers of walls. Doublewall CNTs, which are predominantly found in the samples investigated here, have further advantages over single-wall and multiwall CNTs owing to their coaxial structure. While for multiwall CNTs the inner layers contribute insignificantly to carrying load and reduce their stiffness for a given volume fraction, double-wall CNTs provide an outer wall that can be connected to a composite’s matrix whereas the inner wall remains chemically unaffected and provides the high strength. Thus, the use of double-wall CNTs as a versatile and functionalizable reinforcement material in composites becomes realistic for industrial applications due to the possibility of CNT mass production by CCVD. Acknowledgment. We thank G. Beney (EPFL) for polishing the alumina membranes and Centre Interdisciplinaire de Microscopie Electronique (CIME) at EPFL for access to electron microscopes and technical support. We also acknowledge C. Laurent for interesting discussions. This study is the result of collaboration between the laboratories of the “NanoComp” Training and Mobility program of the European Community. The study in Lausanne was partially supported by the National Center of Competence in Research “Nanoscale science” of the Swiss National Science Foundation. A. Fonseca acknowledges the Region of Wallonia (SYNATEC convention no. 0014526) for financial support. Supporting Information Available: Summary of mechanical measurements. This material is available free of charge via the Internet at http://pubs.acs.org. Nano Lett., Vol. 5, No. 10, 2005
References (1) Iijima, S. Nature 1991, 354, 56-58. (2) Lu, J. P. Phys. ReV. Lett. 1997, 79, 1297-1300. (3) Treacy, M. M. J.; Ebbesen, T. W.; Gibson, J. M. Nature 1996, 381, 678-680. (4) Falvo, M. R.; Clary, G. J.; Taylor, R. M., II; Chi, V.; Brooks, F. P., Jr.; Washburn, S.; Superfine, R. Nature 1997, 389, 582-584. (5) Yu, M. F.; Files, B. S.; Arepalli, S.; Ruoff, R. S. Phys. ReV. Lett. 2000, 84, 5552-5555. (6) Journet, C.; Bernier, P. Appl. Phys. A 1998, 67, 1-9. (7) Wong, E. W.; Sheehan, P. E.; Lieber, C. M. Science 1997, 277, 1971-1975. (8) Dai, H. Top. Appl. Phys. 2001, 80, 29-53. (9) Helveg, S.; Lo´pez-Cartes, C.; Sehested, J.; Hansen, P. L.; Clausen, B. S.; Rostrup-Nielsen, J. R.; Abild-Pedersen, F.; Nørskov, J. K. Nature 2004, 427, 426-429. (10) Salvetat, J.-P.; Kulik, A. J.; Bonard, J.-M.; Briggs, G. A. D.; Sto¨ckli, T.; Me´te´nier, K.; Bonnamy, S.; Be´guin, F.; Burnham, N. A.; Forro´, L. AdV. Mater. 1999, 11, 161-165. (11) Amelinckx, S.; Zhang, X. B.; Bernaerts, D.; Zhang, X. F.; Ivanov, V.; Nagy, J. B. Science 1994, 265, 635-639. (12) Lukic´, B.; Seo, J. W.; Couteau, E.; Lee, K.; Gradecˇak, S.; Berkecz, R.; Hernadi, K.; Delpeux, S.; Cacciaguerra, T.; Be´guin, F.; Fonseca, A.; Nagy, J. B.; Csa´nyi, G.; Kis, A.; Kulik, A. J.; Forro´, L. Appl. Phys. A 2005, 80, 695-700. (13) Amelinckx, S.; Bernaerts, D.; Zhang, X. B.; Van Tendeloo, G.; Van Landuyt, J. Science 1995, 267, 1334-1338. (14) Qian, W.; Liu, T.; Wei, F.; Wang, Z.; Luo, G.; Yu, H.; Li, Z. Carbon 2003, 41, 2613-2617. (15) Babic´, B.; Furer, J.; Sahoo, S.; Farhangfar, Sh.; Scho¨nenberger, C. Nano Lett. 2003, 3, 1577-1580. (16) Bacsa, R. R.; Flahaut, E.; Laurent, Ch.; Peigney, A.; Aloni, S.; Puech, P.; Bacsa, W. S. New J. Phys. 2003, 5, 131. (17) Willems, I.; Konya, Z.; Colomer, J. F.; Van Tendeloo, G.; Nagaraju, N.; Fonseca, A.; Nagy, J. B. Chem. Phys. Lett. 2000, 317, 71-76. (18) The CNTs were prepared by the CCVD method using two protocols. Batch I: The CNTs were prepared by passing a mixture of H2 + CH4 (18% CH4) gas at 1000 °C over nanometer-sized cobalt particles generated in situ. The catalyst particles were produced during the selective reduction of a solid suspension of a Mg1-xCoxO (x ranging from 0.01 to 0.1) powder (see ref 16). The CNTs were annealed at
Nano Lett., Vol. 5, No. 10, 2005
(19)
(20)
(21)
(22) (23) (24)
(25) (26) (27)
600 °C in a graphitization furnace under an argon flow of 3 L/min; a heating rate of 13 °C/min was used before 15 min of resident time at the determined temperature. Batch II: The CNTs were produced by catalytic decomposition of acetylene on an Fe-Mo/NaY zeolite catalyst at 700 °C (see ref 17). The TEM images in Figure 1 were made with a Philips CM300 FEG microscope operating at 300 kV. For the TEM sample preparation, CNTs were dispersed in isopropanol and sonicated for 5 min. A droplet of CNT suspension was then put on a Cu TEM grid with a holey C-film. Salvetat, J.-P.; Briggs, G. A. D.; Bonard, J.-M.; Bacsa, R. R.; Kulik, A. J.; Sto¨ckli, T.; Burnham, N. A.; Forro´, L. Phys. ReV. Lett. 1999, 82, 944-947. The AFM samples were prepared by dispersing the CNT soot in ethanol or chloroform by sonication for 20-30 min. A droplet of the suspension was put on a previously polished alumina membrane (Whatman Anodisc, nominal pore diameter of 200 nm) placed on filter paper and then filtered for few minutes. After solvent evaporation, the CNT bundles were found to occasionally lay over the hole for a short segment of their entire length, with the major part of the bundle still in contact with the membrane surface. AFM images were taken in air with contact mode (Park Scientific Instruments) and at varying loads on each image. Sharpened noncoated or gold-coated Si3N4 cantilevers (ThermoMicroscopes) with a force constant of 0.01-0.1 N m-1 were used. The force constant was calibrated directly prior to the measurement. Gere, J. M.; Timoshenko, S. P. Mechanics of Materials; PWSKENT: Boston, 1990. In the case of an individual nanotube, neglecting the inner diameter leads to the underestimation of the bending modulus by 10%. With an AFM image, one cannot always distinguish between a single nanotube and a bundle. For both batches we obtain EB ) 1 TPa for D < 5 nm. For a bundle, a high value indicates absence of sliding between nanotubes and hence EB ∼ Y, according to eq 1. In the case of a single nanotube, EB ) Y is readily established. Bonard, J.-M.; Chauvin, P.; Klinke, C. Nano Lett. 2002, 2, 665667. Klinke, C.; Bonard, J.-M.; Kern, K. Phys. ReV. B 2005, 71, 035403. Baker, R. T. K. Carbon 1989, 27, 315-323.
NL051034D
2077
