www.advmat.

de
www.MaterialsViews.com

COMMUNICATION
Yarn-Like Carbon Nanotube Fibers
By Juan J. Vilatela and Alan H. Windle*

The covalent bond between carbon atoms will break under a force lubricant. This limitation would also apply to multiwall tubes,
of ∼8 nN, and its stiffness, in terms of force per 100% extension is indeed it has recently been shown by Raman spectroscopy that
some 10 times this at ∼85 nN.[1] For a single graphene sheet, the for double-walled nanotubes stressed in a composite matrix,
breaking force would then be some 32 N/m width, with the stiff- there is very little stress transfer from the outer to the inner
ness some ten times higher. If we consider the graphene sheets layer.[16]
to be stacked at 0.335 nm spacing to make a graphite single The modest levels of the shear strength between the
crystal then we have estimates for the in plane ideal strength of nanotubes endow the fiber with yarn-like characteristics, oth-
125 GPa, and stiffness of 1 TPa. The latter value is measurable, erwise unknown in carbon fibers, where the basic elements are
and in fair agreement with the accepted value of 910 GPa. The nanotubes rather than micrometer-scale discrete sub filaments
tensile strength and stiffness have recently been measured on a as in a textile yarn. The resemblance of CNT fibers to textile
single layer of graphene in the atomic force microscope.[2] The yarns is manifested in the capability of CNT fibers to take
results, 42 N/m and 340 N/m for strength and stiffness respec- exceptional levels of twist.[9,10] In this report we explore aspects
tively, are generally in line with the theoretical values. Values of of their yarn-like characteristics, such as their behavior when
strength and stiffness have also been predicted and measured knotted, cut or under a bending force and relate them to the
for isolated CNTs.[3–6] These however demonstrate the difficul- structure of the CNT yarns. A comparison with conventional
ties of choosing the effective load-bearing cross sectional area, high-performance fibers shows that the unique properties of
an issue also prevalent with fibers in general where the accurate CNT fibers derive from its semi-continuous structure.
measurement of diameter and thus cross sectional area is often The fiber is made from an aerogel consisting of an open net-
challenging. For this reason, we adopt N/tex as the unit of stress work of nanotube bundles, which is drawn down and wound
for all axial properties, which is exactly numerically equivalent up.[9] If the fiber is not densified on line, it winds as a very thin
to the stress/specific gravity, known generally as specific stress. film (Figure 1a). Spraying acetone onto it, which then rapidly
The relationship is explained in more detail in the Supporting evaporates, causes the film to condense into a fiber of approxi-
Information, Section SOM 1. The value of specific stress is not mately circular cross section (Figure 1b).[15] The fact that the
dependent on the lateral packing of the graphene layers as long film consists of aligned bundles of nanotubes endows it with
as they all contain the common fiber axis. They could be single yarn-like properties. So although the fiber is a filament some
wall tubes, multiwall tubes, collapsed tubes, scrolls, turbostrati-
cally stacked sheets, or sheets packed in crystallographic register
as in a crystal.
The exploitation of the highly desirable axial properties of
nanotubes in fibers depends on aligning the carbon nanotubes
parallel to the fiber, as with the alignment of chain molecules
with the fiber axis in polymer fibers. There are several proc-
esses by which such alignment has been achieved.[7–10] While
reported values of strength and stiffness are in the range of
other types of high-performance fibers, their tensile properties
(particularly for gauge lengths longer than a few millimeters)
still fall short of those of their constituent nanotubes.[8,10–15]
Strength comparisons between differently made carbon nano-
tube fibers and also with other performance fibers are shown
graphically in reference.[15] The factor which is central in
limiting strength is the difficulty in transferring stress from
one nanotube of finite length to a neighbor by shear across
the interface, as the shear strength of the interface between
graphene layers is very low–graphite is, after all, used as a

Figure 1. (a) SEM image of an oriented thin film of carbon nanotubes

before condensation into a fiber. The fibrillar features are bundles of
[∗] J. J. Vilatela, Prof. A. H. Windle carbon nanotubes and the bright spots catalyst particle impurities.
Department of Materials Science (b) SEM image of the fiber after condensation by acetone vapor (c) SEM
Pembroke Street, Cambridge, CB2 3QZ (UK) image of a CNT fiber pressed onto the edge of a razor blade (vertical line).
E-mail: [email protected] (Image courtesy of K. Koziol) (d) SEM image of a CNT fiber showing a
kink induced by bending, and the modification of the cross section at
DOI: 10.1002/adma.201002131 that point.

Adv. Mater. 2010, 22, 4959–4963 © 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 4959
 www.advmat.de
www.MaterialsViews.com
COMMUNICATION

propagation as the crack is constantly diverted laterally where it

meets the weak interfaces between the nanotube bundles.
A second aspect is the extreme ease of bending. The com-
paratively weak shear strength between the fiber’s compo-
nents means that in the relationship between bending stiff-
ness (Eb) and diameter (d) (Eb α d4), d is probably closer to
100 nm than to the diameter of the fiber (10 µm). Due to the
weak interaction between the sub-elements in the fiber, it
also flattens in cross section at the point where the bending
concentrates into a kink (Figure 1d), effectively decreasing
the fiber thickness at that point, and providing even lower
bending stiffness which now scales as t2. Also, the nanotube
bundles on the compressive side of the fiber buckle without
apparent damage to the fibers, with the result that the neu-
tral plane moves outwards reducing the tension on the out-
side of the fiber, as observed in conventional yarns subject to
bending.[17] When the bending stress is released the fiber can
be straightened again with a very small axial stress (<0.001 N/tex)
and there is no apparent degradation of tensile properties,
although the cross section remains flattened, as shown in
Figure 1d.
Perhaps the clearest demonstration of yarn-like nature is
the mode of tensile fracture of the fiber. Figure 2 shows both
parts of a fiber fracture. It is fibrous in nature and demon-
strates not only a pulling out of fiber bundles, but also that
the bundles form an apparently continuous network as they
bifurcate and then rejoin. The mode of fracture clearly sug-
gests that the strength is controlled by a combination of inter-
face strength in shear and the length of the internal fibrous
elements.
The yarn-like nature of this fiber is illustrated not only
by its response to elastic bending as described above, but
by its performance in the so-called “knot test”. In this test
the fiber is tested conventionally in tension and then again
after a simple overhand knot has been tied in it (Figure 3).

Figure 2. SEM images showing the fracture of a CNT fiber. The images
indicate that failure through shear between the fibrous elements is an
important aspect of the fracture of the yarn-like fiber.

10 micrometers in diameter, it is also a yarn made up of

component filaments some 100 nm in diameter, each of these
consisting of ∼100 carbon nanotubes. It can be reasonably
described as a yarn, or at least a proto-yarn. One aspect is its
behavior when it is cut. Figure 1c shows a carbon nanotube
fiber stretched across the cutting edge of a razor blade (vertical
line) where it has spread laterally, a mechanism which pro- Figure 3. SEM image of a CNT fiber with an overhand knot. (Inset) Note
vides resistance to cutting resulting from the difficulty of crack that the knotted section of the fiber is roughly 24 µm long but only 6 µm
wide, showing the flattening of the fiber.

4960 wileyonlinelibrary.com © 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 4959–4963
 www.advmat.de
www.MaterialsViews.com

COMMUNICATION
Figure 4. Strength histograms for: (a) CNT fiber, (b) Kevlar 49 (c) Dyneema and (d) T300 carbon fiber with and without an overhand knot. All measure-
ments were carried out under identical testing conditions.

The knot efficiency is the ratio of the strength of a knotted The nature of the knot in CNT fibers is unusual, mainly
fiber to that of an unknotted control expressed as a per- stemming from the ease of bending this proto-yarn. The knot
centage. Figure 4 shows the strength distributions measured tightens to dimensions not very much greater than those of
for knotted and unknotted examples of CNT fiber, compared the fiber itself with surprisingly little load. Figure 5c shows the
with measurements made under identical conditions for knot loop diameter plotted against tensile stress as the knot is
aramid (Kevlar 49), polyethylene (Dyneema) and Carbon Fiber tightened in the testing machine. For the CNT fiber the knot
(T300). The linear density and diameter of the fibers tested loop reduces in diameter to 100 µm or less before any detect-
is included in the Supporting Information, Section SOM 2. able tensile stress is required, while Kevlar and Dyneema fibers
The reason why we have not managed to obtain more data for require in excess of 0.02 N/tex before they are tightened to the
knotted examples of T300 carbon fiber is that the fiber tended same extent. This is indicative of the low bending modulus of
to break before the knot could be properly established, even at the CNT fiber.
a large initial loop radius. All values of knot strength quoted The yarn-like nature of this carbon nanotube fiber, in ena-
here were measured in our laboratory under identical test bling very easy bending, appears to provide knot efficiencies
conditions. not normally seen in a high performance fiber, with absolute
Remarkably, carbon nanotube fibers showed knot efficiencies knot strength exceeding those of any other fiber except for
of 100%, demonstrating no degradation of strength as a result Dyneema.
of the knot, with most fibers breaking remote from the knot The particular yarn-like properties of carbon nanotube fiber
itself. The comparison with a wide range of fibers is shown in made by direct spinning make it stand out against all conven-
Figure 5a, which is a plot of knot strength efficiency against tional carbon fibers with respect to its ability to bend through
strength. The trend is similar for that of knot strength against very tight radii without apparent permanent damage. Because
stiffness (Figure 5b). the structure of the fiber itself consist of a network of smaller

Adv. Mater. 2010, 22, 4959–4963 © 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 4961
 www.advmat.de
www.MaterialsViews.com
COMMUNICATION

fiber and as a result compromise to some extent the 100% knot

efficiency seen here, the essential yarn-like nature of the fiber
is expected to maintain its dominance with respect to this par-
ticular combination of properties.

Experimental Section
The CNT fibers were produced by continuously drawing the CNT aerogel
from the gas phase in the reactor.[9] The nanotubes in the aerogel were
synthesized at 1250 °C by chemical vapor deposition of a mixture
containing, ferrocene, thiophene and a carbon source, with hydrogen
as a carrier gas. They consisted mostly of single and double-walled
nanotubes, some of them being collapsed due to their large diameter.[18]
The CNT fiber was densified on-line using a mist of atomized acetone,
and wound at a rate of 20–30 m/min.
Testing of fiber was carried out with a standard instrument used in the
fiber industry, a Textechno tensile tester. The linear density of the fibers
was determined by the acoustic method also using this instrument,
and by directly weighing known lengths of fiber on a high resolution
Sartorius balance. The fiber gauge length was kept at 20 mm for all tests.
Overhand knots were tied in the fiber using metal tweezers with the aid
of a stereo microscope. Besides CNT fibers, the following fibers were
also tested in-house: Dyneema, M5, T300, Kevlar 29, Kevlar 49, PBO,
K1100, Twaron and Vectran. The values for the rest of the fibers shown
in Figure 5a and 5b were obtained from reference 17.
SEM images were obtained using a JEOL 6340F FEGSEM and a JEOL
820 SEM.
The tightening of the fiber knot was recorded with a Canon XM1 with
a supplementary close-up lens on the front. Knot loop diameters were
determined by image analysis of the video.

Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.

Acknowledgements
The authors wish to thank Professor J.W.S Hearle for a most helpful
discussion about yarns and the US Army International Technology Center
London, CONACyT Mexico and the EPSRC for funding.

Received: May 17, 2010

Revised: July 9, 2010
Published online: August 31, 2010

[1] K. E. Perepelkin, Mekhan. Polim. 1966 1, 34

Figure 5. (a) Plot for a variety of fibers of knot strength efficiency against [2] C. Lee, X. Wei, J.W. Kysar, J. Hone, Science 2008 321, 385.
fiber strength for a variety of fibers, (b) Similar plot against fiber stiffness. [3] B. I. Yakobson, C. J. Brabec, J. Bernholc, Phys. Rev. Lett. 1996 76,
(c) Plot showing the relationship between the diameter of the knot loop 2511.
and the applied tensile stress. Note that in the case of CNT fiber the loop [4] J. Lu, Phys. Rev. Lett. 1997 79, 1297.
tightens with very little detectable tensile load. [5] M. Yu, O. Lourie, M. J. Dyer, K. Moloni, T. F. Kelly, R. S. Ruoff,
Science 2000 287, 637.
filament subunits (nanotubes and bundles), it can tolerate [6] B. Peng, M. Locascio, P. Zapol, S. Li, S. L. Mielke, G. C. Schatz,
bending in ways that a traditional fiber with a continuous H. D. Espinosa, Nat. Nanotechnol. 2008 3, 626.
[7] B. Vigolo, A. Pénicaud, C. Coulon, C. Sauder, R. Pailler, C. Journet,
solid structure cannot. This type of behavior is demonstrated
P. Bernier, P. Poulin, Science 2000 290, 1331.
by the fiber’s performance in a knot test, where the particular [8] L. M. Ericson, H. Fan, H. Peng, V. A. Davis, W. Zhou, J. Sulpizio,
combination of strength, stiffness and knot strength obtained Y. Wang, R. Booker, J. Vavro, C. Guthy, A. N. G. Parra-Vasquez,
appear unrivaled by any other fiber currently available (Figure 5). M. J. Kim, S. Ramesh, R. K. Saini, C. Kittrell, G. Lavin, H. Schmidt,
While a range of post-spinning treatments, becoming available W. W. Adams, W. E. Billups, M. Pasquali, W. Hwang, R. H. Hauge,
now, may well increase the stiffness and strength of the CNT J. E. Fischer, R. E. Smalley, Science 2004 305, 1447.

4962 wileyonlinelibrary.com © 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010, 22, 4959–4963
 www.advmat.de
www.MaterialsViews.com

COMMUNICATION
[9] Y. Li, I. A. Kinloch, A. H. Windle. Science 2004 304, 276. [14] P. Miaudet, S. Badaire, M. Maugey, A. Derre, V. Pichot, P. Launois,
[10] M. Zhang, K. R. Atkinson, R. H. Baughman. Science 2004 306, 1358. P. Poulin, C. Zakri, Nano Lett. 2005 5, 2212.
[11] X. Zhang Q. Li, T. G. Holesinger, P. N. Arendt, J. Huang, [15] K. Koziol, J. Vilatela, A. Moisala, M. Motta, P. Cunniff, M. Sennett,
P. D. Kirven, T. G. Clapp, R. F. DePaula, X. Liao, Y. Zhao, L. Zheng, A. Windle, Science 2007 318, 1892.
D. E. Peterson, Y. Zhu, Adv. Mat. 2007, 19, 4198. [16] S. Cui, I. A. Kinloch, R. J. Young, L. Noe, M. Monthioux, Adv. Mater.
[12] S. Zhang, L. Zhu, M. L. Minus, H. G. Chae, S. Jagannathan, C. 2009 21, 3591.
Wong, J. Kowalik, L. B. Roberson, S. Kumar, J. Mater. Sci. 2008 43, [17] W. W. Morton, J. W. Hearle, Physical properties of textile fibers, 4th ed.
4356. (The textile Institute, Manchester, 2008)
[13] S. Zhang, K. Koziol, I. A. Kinloch, A. H. Windle, Small 2008 4, [18] M. S. Motta, A. Moisala, I. A. Kinloch, A. H. Windle, Adv. Mater.
1217. 2007 19, 3721.

Adv. Mater. 2010, 22, 4959–4963 © 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 4963