Accepted Manuscript Title: In-situ observation of carbon nanotube yarn during voltage application Author: Tomoharu Tokunaga Yasuhiko Hayashi Toru Iijima Yuki Uesugi Masaki Unten Katsuhiro Sasaki Takahisa Yamamoto PII: DOI: Reference:
S0968-4328(15)00054-2 http://dx.doi.org/doi:10.1016/j.micron.2015.04.004 JMIC 2182
To appear in:
Micron
Received date: Revised date: Accepted date:
12-3-2015 9-4-2015 9-4-2015
Please cite this article as: Tokunaga, T., Hayashi, Y., Iijima, T., Uesugi, Y., Unten, M., Sasaki, K., Yamamoto, T.,In-situ observation of carbon nanotube yarn during voltage application, Micron (2015), http://dx.doi.org/10.1016/j.micron.2015.04.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ip t
In-situ observation of carbon nanotube yarn during voltage application
Tomoharu Tokunaga1*, Yasuhiko Hayashi2, Toru Iijima3, Yuki Uesugi2, Masaki Unten2,
us
Department of Quantum Engineerign, Nagoya University, Furo-cho, Chikusa-ku,
an
1
Nagoya, Aichi, 464-8603, Japan
Department of Electric Engineering, Okayama University, Tsushimanaka, 3-1-1,
Environment Energy Nanotechnology Research Institute Co., Ltd, Aizome 5263-5,
te
3
d
Kita-ku, Okayama, 700-8530, Japan
M
2
cr
Katsuhiro Sasaki1, Takahisa Yamamoto1
Ac ce p
Ikeda-cho, Kitaazumi-gun, Nagano, 399-8602, Japan
Corresponding author: Tomoharu Tokunaga Tel: +81-52-789-3350, E-mail: [email protected]
Page 1 of 19
Abstract
ip t
Carbon nanotube (CNT) yarns are fabricated by drawing (combined with spinning) from CNT forests and grown on a substrate. Three types of phenomena occur in these
cr
CNT yarns with increasing amounts of current: yarn rotation, catalyst evaporation, and
us
breakage of the yarn. These phenomena result from the resistive heating occurring
an
during the current flow, and have been observed in situ under vacuum by transmission electron microscopy. If these CNT yarns are applied to electronic circuits, the rotation
M
and breakage may lead to circuit failure. However, catalyst evaporation is a useful
Ac ce p
te
d
method for purifying CNT yarns without additional treatments prior to yarn fabrication.
Page 2 of 19
Introduction
ip t
Previous research has shown that carbon nanotubes (CNTs) have a range of unique, superior properties, including electrical conductivity, thermal conductivity, and
cr
mechanical strength (Backtold et al., 1999, Zou et al., 2004, Wong et al., 1997, Meo and
us
Rossi, 2006). The combination of these properties and their nanoscale structure has led
an
to a variety of applications that utilize CNTs. While previous studies reported on thin films of CNTs, recent studies have reported on bundling CNTs to form yarns and have
M
reported their electrical and mechanical properties. CNT yarns are fabricated using a
d
parallel drawing and spinning method. Twisted CNT yarns are produced directly from
te
substrates where the vertically aligned CNTs were grown (Jiang et al., 2002, Zhang et
Ac ce p
al., 2004, Tran et al., 2009). The tensile modulus of CNT yarn increases concurrently with its diameter, with a maximum strength of ~30 GPa at a diameter of 2 m (Zhang et al., 2004). Due to the high strength of CNT yarns, it has been suggested that these yarns could possibly be used to form cables to transport people and items between the earth and space as part of a space elevator. Theoretically, the conductivity of a CNT yarn can be reduced to that of a single CNT for use as a conducting wire (Miao, 2011). However, the physical phenomena responsible for the current flow in CNT yarns have not yet been reported. An understanding of these phenomena is vital to increase the
Page 3 of 19
applicability of CNT yarns, including using the yarns for conducting wires. In this study,
ip t
CNT yarns with and without current flow were observed and analyzed by transmission electron microscopy (TEM) and Raman spectra in order to develop a better
us
cr
understanding of this phenomena.
an
Materials and Methods
For the current flow analysis, CNT yarns were fabricated by a combined drawing with
M
spinning method from a dense, tall, and highly vertically aligned CNT forest grown on a
d
substrate. First, the CNTs were grown using a thermal chemical vapor deposition
te
procedure (CVD, BlackMagic II, Germany, Aixtron). A 2-nm layer of iron was
Ac ce p
deposited onto a Si substrate as a precursor, and then CNTs were grown on this substrate in an acetylene atmosphere via thermal CVD. The resulting CNTs are ~300 m in length, 6 nm in diameter, and approximately three carbon walls thick, as previously shown (Iijima et al, 2011). A section of the grown CNTs near the edge of the substrate was taken and drawn in parallel to the substrate surface while spinning. For the CNT yarn fabrication, the drawing speed and spinning rate were 40 mm/min and 4000 rpm, respectively. The fabricated CNT yarn was rolled on a paper rod (Fig. 1), and its structure was then observed via in situ TEM analysis.
Page 4 of 19
paper rod
an
Figure 1: Photograph of the drawn CNT yarns
us
3 mm
cr
ip t
CNT yarn
M
During current flow, the CNT yarn was observed and recorded by a digital camera
d
system (SC200D, Gatan, California, USA). A high-voltage TEM, equipped with a
te
digital camera system for in situ observation, is used for visual measurements (HVTEM,
Ac ce p
JEM-1000K RS, Akishima, Japan, JEOL). The CNT yarn was connected between two of the four electrodes equipped on the TEM holder. Typically, samples connected between these electrodes are fixed with screws. However, the CNT yarn diameter is too small to be fixed with a screw; and therefore, it was connected to the electrodes with a conducting paste (DOTITE D-550, Fujikura Kasei, Tokyo, Japan). After insertion into the TEM chamber, current
passed from the electrode through an electrical wire from
the TEM holder to outside of the chamber. A variable resistor controlled the current from a battery,
the electrodes carried the current to the CNT yarn. An electron energy
Page 5 of 19
loss spectroscope (EELS: Quantum, California, USA, Gatan) was used for elemental
current
ip t
analysis. Raman spectra were acquired for CNT yarns with and without exposure to to identify their crystallization. The Raman spectrometer was equipped with a
cr
532-nm wavelength laser (NRS-1000VUV, Tokyo, Japan, Jasco).
us
From TEM observations, the CNT yarn has a diameter of 5 m and some individual
an
CNTs protrude from the surface of yarn (Fig 2(a)). This suggests the yarn is not perfectly spun. Moreover, a closer examination of the edges of the yarn revealed spaces
M
between the CNTs (Fig. 2(b)). These spaces were further examined via electron energy
d
loss spectroscopy (EELS) (inset of Fig. 2(b)). The EELS spectra of the small dark
te
circular contrasts confirmed the presence of Fe-L3, L2, and O-K edges. Based on these
Ac ce p
results, it was determined that the circular contrasts are iron oxide particles. The ratio of the Fe-L3 and L2 intensity (L3/L2), which was measured from the deconvoluted Fe-L3 and L2 EELS spectra, gives the iron valence of iron oxide (Sparow et al., 1984). The L3/L2 ratio of iron oxide particle is ~3.3, suggesting the iron oxide particle is FeO. These FeO particles originate from the oxidized iron catalyst particles used for the CNT growth. Prior to the CNT growth, the iron catalyst is annealed after deposition by evaporation on the substrate surface. Our results show that the iron catalyst is expected to oxidize during the sample transfer and/or the fabrication of the CNT yarn.
Page 6 of 19
(b)
us
cr
ip t
(a)
200 nm
an
2 m
M
Figure 2: TEM images of (a) the center and (b) near the CNT yarn edge, inside the
d
dotted black line rectangle of Figure 2(a). The inset in Figure 2(b) shows the EELS
Ac ce p
te
acquired from inside the dotted line circle.
Results
Low-magnification TEM images of the CNT yarn over time are observed during a 5
mA current flow through the yarn containing FeO particles (Fig. 3). The centers of the TEM images contain thin, long contrast areas in which the black small catalyst particles are apparent. Only particles with diameters greater than ~250 nm are observed in the CNT yarn using low-magnification TEM During current flow, the large particles repeatedly move left and right on the yarn. In addition, both the imperfectly spun CNTs
Page 7 of 19
and the larger particles on the CNT yarn are observed to rotate around the center axis of
ip t
the yarn. Therefore, the CNT yarn is confirmed to rotate during current flow. This rotation phenomenon is not continuous; after one revolution, the rotation nearly stops.
cr
Further, when currents of 5 mA were applied to the yarn, rotation became negligible. As
us
a conductive past fixes both ends of the CNT yarn in place, the sole driving force for the
15s
30s
45s
te
75s
90s
Ac ce p
60s
d
M
0s
an
rotational phenomena is the current flow.
Figure 3: In situ TEM images of the CNT yarn over time during 5-mA current flow.
Next, the CNT yarn with a high current (26 mA) flow is observed through time
progressive TEM images (Fig. 4). The catalyst particles are illustrated in the small black circular area marked by white arrows. These catalyst particles decrease in size with time before completely disappearing. When current is applied to the yarn, its temperature
Page 8 of 19
rises past the melting point (MP) of FeO (1369 °C) via resistive heating (Robert and
ip t
Weast, 1973). Thus, the catalyst particles are expected to evaporate from the yarn. As the CNT yarn is not perfectly spun, the evaporated particles could escape from
Although the catalyst particles evaporated from the
cr
numerous pores within the yarn.
us
CNT yarn, its shape and diameter are unaffected. Due to similarities with graphene
significantly higher than the MP of FeO.
an
sheets, the MPs of the CNTs and graphene are approximately identical, which is Based on the evaporation of the catalysts, it
Ac ce p
te
d
M
is expected that current flow leads to the resistive heating of the CNT yarn.
Page 9 of 19
Figure 4: In situ TEM images of the CNT yarn over time during 26-mA current flow.
ip t
When a 31 mA current is applied to the CNT yarn, the CNT yarn disappears from observation. The TEM images show that the CNT yarn is cut from the holder (Fig. 5).
cr
The diameter of the cut end of the yarn is smaller than the diameter of the remaining
Furthermore, regions of the cut CNT yarn exhibit some amorphous carbon,
an
5(b)).
us
CNT yarn (Fig. 5(a)). The cut end is very small, consisting of only a few CNTs (Fig.
indicating that the CNT yarn sublimed from the resistive heating due to the large current
M
flow. Since the heat capacity of the electrodes is considerably higher than the yarn, the
d
CNT yarn heated considerably during the current flow, yet the temperature close to the
te
electrodes did not rise. The center of the CNT yarn between the electrodes heated to the
Ac ce p
highest temperature. Thus, the CNT sublimed in this region from the resistive heating and deposited on the remaining CNT yarn that was at a lower temperature. Furthermore, the diameter of the CNT yarn gradually decreases approaching the broken end of the yarn, due to the current flowing through the surface.
For the region of yarn with the
greatest degree of heating, there is a large surface area with a low current density prior to CNT sublimation through resistive heating. As the CNT yarn is heated, the CNT sublimation occurs over a large area with a high current flow. As the CNTs sublimates on the surface of the yarn, the surface area decreases and the current density increases.
Page 10 of 19
The increase in current density leads to further CNT sublimation around the center of
diameter of the cut end of the CNT yarn is expected to decrease.
(b)
M
an
us
cr
(a)
ip t
the reduced surface area. As the CNT sublimation increases at the highest heat, the
5 m
te
d
50 nm
Ac ce p
Figure 5: In situ TEM images of the cut CNT yarn after 31-mA current flow.
The Raman spectrum is acquired for the initial CNT yarn prior to current flow and the
cut CNT yarn (Fig. 6). Two peaks appear in each spectrum, with the D-band peak at ~1350 cm−1 and the G-Band peak at ~1580 cm−1. The G-band peak results from vibrations on the 6-membered ring of carbon atoms along the nanotube axis and circumference (Dresselhaus et al., 2005). The D-band peak results from amorphous or disordered carbon and defects in the CNT samples (Datshuk et al., 2008). Typically, the
Page 11 of 19
intensity ratio of the G-band to the D-band peaks (IG/ID) is used to determine the CNT
ip t
purity and crystallization. CNTs with a high IG/ID ration are purer with a higher degree of crystallinity than those with a low IG/ID ratio. The IG/ID ratio of the initial CNT yarn
cr
is 0.90, while that of the cut yarn is 1.52 (Fig. 6) As the CNT yarn temperature increases
us
due to resistive heating during current flow, the crystallization of individual CNTs
an
within the yarn improves due to high-temperature exposure, as no amorphous carbon is observed in the initial CNT yarn (Fig. 2). In general, the CNT crystallization improves
M
with heating over 2000 ºC under vacuum (Ci et al., 2001, Huang et al., 2004). The
d
progression of crystallization is not apparent from our observations, but the effect of the
Ac ce p
te
raised temperature on the CNT yarn is apparent in the Raman spectra.
Page 12 of 19
Figure 6: Raman spectra acquired from the CNT yarn before (initial) and after current
ip t
flow (of the cut CNT yarn).
cr
Discussion
us
The above results demonstrate an important phenomena occurring with the CNT yarn
an
during current flow. As the current flows through the CNT yarn, its temperature rises, and the behavior the yarn changes with increasing temperature. Initially, a rotation of
M
the CNT yarn is observed. This rotation phenomenon is expected based on the yarn’s
d
thermal expansion. The CNT yarn elongated with heating, and it is expected that the
te
yarn would expand along its diameter. However, with the yarn length in millimeters and
Ac ce p
diameter in micrometers, the thermal expansion coefficients along its length and diameter directions are nearly equal. Since the elongation of the CNT yarn is more apparent than its expansion, the thermal elongation must be considered when discussing the rotation phenomenon. The twist density of the CNT yarn decreases with elongation through thermal expansion. Therefore, the rotation phenomena results from twist unraveling. For the length of the CNT, the thermal expansion coefficient increases with temperature, becoming nearly saturated at ~1300 ºC (Jiang et al., 2004). Above 1300 ºC, the coefficient is essentially constant, ceasing the rotation phenomena. While nearly all
Page 13 of 19
materials elongate with heating, this rotation is distinctive of CNT yarns. This rotation
ip t
phenomenon of the CNT yarn may affect its application in electric circuits. If the temperature of the CNT yarn is not controlled when used as an electric wire, it will
cr
elongate and rotate. This could lead to the CNT yarn moving and interfering with the
us
surrounding electrical elements, which may result in the failure of the electrical circuit.
drawn CNT yarn may not be continuous.
an
The spinning of the yarn can be removed from the fabrication process, although the
M
The catalyst evaporation phenomena observed after the rotation is an important
d
property of CNT yarn. Previously, metal catalysts were removed by heating the CNT
te
under vacuum or exposing it to an acid solution. These purified CNTs were then drawn
Ac ce p
to obtain a CNT yarn without the metal catalyst. However, our results reveal that it is possible to obtain a CNT yarn without metal catalysts by applying a current during the drawing process. This would allow for electrical circuit fabrications with a CNT yarn. Finally, at large current flows, it is shown that the CNT yarn sublimates, breaking the
yarn. The sublimated carbon is deposited on the nearby, cooler areas of yarn, and it is amorphous with low resistivity. If incorporated, this could affect the properties of an electronic circuit if the CNT yarn (Miyagawa et al., 2006). However, the carbon
Page 14 of 19
sublimation temperature is at ~3550 ºC (Robert and Weast, 1973). Thus, the CNT yarn
ip t
is suitable for use as an electric wire in high-temperature electronic devices.
cr
Conclusions
us
During the application of different current levels, in situ TEM observations of CNT
an
yarns with oxidized catalyst particles revealed yarn rotation, catalyst evaporation, and a cutting phenomenon. It is believe that these phenomena result from resistive heating
M
during the current flow through the yarn. Moreover, it is clear that CNT yarns could be
d
used as electrical wires at high temperature, around carbon sublimation temperature,
te
and under vacuum. These phenomena need to be recognized when incorporating CNT
Ac ce p
yarns into novel electronic devices, particularly as wires. These phenomena could both positively and adversely affect the device.
Acknowledgements
This work was supported by the Advanced Characterization Nanotechnology Platform
of the National Institute for Materials Science and the High Voltage Electron Microscopy Laboratory of Nagoya University.
Page 15 of 19
References
ip t
Iijima S., 1991. Helical microtubules of graphitic carbon. Nature. 354, 56-58. Bachtold, A., Strunk, C., Salvetat, P.J., Bonard, M.J., Forró, L., Nussbaumer, T.,
cr
Schönenberger, C., 1999. Aharonov-Bohm oscillations in carbon nanotubes. Nature.
us
397, 673-675.
an
Zou, P.X., Abe, H., Shimizu, T., Ando, A., Nakayama, Y., Tokumoto, H., Zhu, M.S., Zhout, S.H., 2004. Simple thermal chemical vapor deposition synthesis and
M
electrical property of multi-walled carbon nanotubes. Physica E. 24, 14-18.
d
Wong, W.E., Sheehan, E.P., Liber, M.C., 1997. Nanobeam Mechanics. Elasticity,
te
Strength, and Toughness of Nanorods and Nanotubes. Science. 277, 1971-1975.
Ac ce p
Meo, M., Rossi, M., 2006. Tensile failure prediction of single wall carbon nanotube. Eng. Fract. Mechanics. 73, 2589-2599.
Jiang, K., Li, Q., Fan, S., 2002. Spinning continuous carbon nanotube yarns. Nature. 419, 801.
Zhang, M., Atkinson, R.K., Baughman, H.R., 2004. Multifunctional Carbon Nanotube Yarns by Downsizing an Ancient Technology. Science. 306, 1358-1361.
Page 16 of 19
Tran, D.C., Humphries, W., Smith, M.S., Huynh, C., Lucas, S., 2009. Improving the
ip t
tensile strenghth of carbon nanotube spun yarns using a modified spinning process. Carbon. 47, 2662-2670.
cr
Miao, M., 2011. Electrical conductivity of pure carbon nanotube yarns. Carbon. 49,
us
3755-3761.
an
Iijima, T., Oshima, H., Hayashi. Y., Suryavanshi, B.U., Hayashi, A., Tanemura, M., 2011. Morphology control of a rapidly grown vertically aligned carbon-nanotube
M
forest for yarn spinning. Phys. Stat. Sol. A. 10, 2332-2334.
te
CRC PRESS, Ohio
d
Robert, C., Weast, P.D., 1973. CRC Handbook of chemistry and physics 54th edition.
Ac ce p
Sparow T.G., Williams B.G., Rao C.N.R., Thomas J.M., 1984. L3/L2 WHITE-LINE INTENSITY RATIOS IN THE ELECTRON ENERGY-LOSS SPECTRA OF 3d TRANSITION-METAL OXIDES. Chem. Phys. Lett. 108, 547-550.
Dresselhaus, S.M., Dresselhaus, G., Saito, R., Jolio, A., 2005. Raman spectroscopy of carbon nanotubes. Phys. Rep. 409, 47-99.
Datshuk, V., Kalyva, M., Papagelis, K.P., Tasis, D., Siokou, A., Kallitsis, I., Galiotis, C., 2008. Chemical oxidation of multiwalled carbon nanotubes. Carbon. 46, 833-840.
Page 17 of 19
Ci, L., Wei, B., Xu, C., Liang, J., Wu, D., Xie, S., Zhou, W., Li, Y., Liu, Z., Tang, D.,
ip t
2001. Crystallizationh behavior of the amorphous carbon nanotubes prepared by the CVD method. J. Crystal Growth. 233, 823-828.
cr
Huang, W., Luo, G., Wei, F., 2003. 99.9% purity multi-walled carbon nanotubes by
us
vacuum high-temperature annealing. Carbon. 41, 2585-2590.
an
Jinag, H., Liu, B., Huang, Y., Hwang, C.K., 2004. Thermal Expansion of Single Wall Carbon Nanotubes. J. Eng. Mater. Tech. 126, 265-270.
M
Iijima, T., Oshima. H., Hayashi, Y., Suryavanshi, B.U., Hayashi, A., Tanemura, M.,
d
2012. In-situ observation of carbon nanotube yarn from vertically aligned carbon
te
nanotube forest. Diam. Rel. Mater. 24, 158-160.
Ac ce p
Miyagawa, S., Nakao, S., Choi, J., Ikeyama, M., Miyagawa, Y., 2006. Effects of target bias voltage on the electrical conductivity of DLC films deposited by PBII/D with a bipolar pulse. Nucl Inst. Meth. in Phys. Res. B. 242, 346-348.
Robert, C., Weast, P.D., 1973. CRC Handbook of chemistry and physics 54th edition. CRC PRESS, Ohio.
Page 18 of 19
Highlights
ip t
The CNT yarn during current flow was observed by TEM. After starting current flow, rotation and cutting phenomena on the CNT yarn were
cr
confirmed.
us
Moreover, catalyst, which was used for CNT growth, was evaporated.
Ac ce p
te
d
M
an
These phenomena occur due to resistive heating during current flow.
Page 19 of 19
S0968-4328(15)00054-2 http://dx.doi.org/doi:10.1016/j.micron.2015.04.004 JMIC 2182
To appear in:
Micron
Received date: Revised date: Accepted date:
12-3-2015 9-4-2015 9-4-2015
Please cite this article as: Tokunaga, T., Hayashi, Y., Iijima, T., Uesugi, Y., Unten, M., Sasaki, K., Yamamoto, T.,In-situ observation of carbon nanotube yarn during voltage application, Micron (2015), http://dx.doi.org/10.1016/j.micron.2015.04.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ip t
In-situ observation of carbon nanotube yarn during voltage application
Tomoharu Tokunaga1*, Yasuhiko Hayashi2, Toru Iijima3, Yuki Uesugi2, Masaki Unten2,
us
Department of Quantum Engineerign, Nagoya University, Furo-cho, Chikusa-ku,
an
1
Nagoya, Aichi, 464-8603, Japan
Department of Electric Engineering, Okayama University, Tsushimanaka, 3-1-1,
Environment Energy Nanotechnology Research Institute Co., Ltd, Aizome 5263-5,
te
3
d
Kita-ku, Okayama, 700-8530, Japan
M
2
cr
Katsuhiro Sasaki1, Takahisa Yamamoto1
Ac ce p
Ikeda-cho, Kitaazumi-gun, Nagano, 399-8602, Japan
Corresponding author: Tomoharu Tokunaga Tel: +81-52-789-3350, E-mail: [email protected]
Page 1 of 19
Abstract
ip t
Carbon nanotube (CNT) yarns are fabricated by drawing (combined with spinning) from CNT forests and grown on a substrate. Three types of phenomena occur in these
cr
CNT yarns with increasing amounts of current: yarn rotation, catalyst evaporation, and
us
breakage of the yarn. These phenomena result from the resistive heating occurring
an
during the current flow, and have been observed in situ under vacuum by transmission electron microscopy. If these CNT yarns are applied to electronic circuits, the rotation
M
and breakage may lead to circuit failure. However, catalyst evaporation is a useful
Ac ce p
te
d
method for purifying CNT yarns without additional treatments prior to yarn fabrication.
Page 2 of 19
Introduction
ip t
Previous research has shown that carbon nanotubes (CNTs) have a range of unique, superior properties, including electrical conductivity, thermal conductivity, and
cr
mechanical strength (Backtold et al., 1999, Zou et al., 2004, Wong et al., 1997, Meo and
us
Rossi, 2006). The combination of these properties and their nanoscale structure has led
an
to a variety of applications that utilize CNTs. While previous studies reported on thin films of CNTs, recent studies have reported on bundling CNTs to form yarns and have
M
reported their electrical and mechanical properties. CNT yarns are fabricated using a
d
parallel drawing and spinning method. Twisted CNT yarns are produced directly from
te
substrates where the vertically aligned CNTs were grown (Jiang et al., 2002, Zhang et
Ac ce p
al., 2004, Tran et al., 2009). The tensile modulus of CNT yarn increases concurrently with its diameter, with a maximum strength of ~30 GPa at a diameter of 2 m (Zhang et al., 2004). Due to the high strength of CNT yarns, it has been suggested that these yarns could possibly be used to form cables to transport people and items between the earth and space as part of a space elevator. Theoretically, the conductivity of a CNT yarn can be reduced to that of a single CNT for use as a conducting wire (Miao, 2011). However, the physical phenomena responsible for the current flow in CNT yarns have not yet been reported. An understanding of these phenomena is vital to increase the
Page 3 of 19
applicability of CNT yarns, including using the yarns for conducting wires. In this study,
ip t
CNT yarns with and without current flow were observed and analyzed by transmission electron microscopy (TEM) and Raman spectra in order to develop a better
us
cr
understanding of this phenomena.
an
Materials and Methods
For the current flow analysis, CNT yarns were fabricated by a combined drawing with
M
spinning method from a dense, tall, and highly vertically aligned CNT forest grown on a
d
substrate. First, the CNTs were grown using a thermal chemical vapor deposition
te
procedure (CVD, BlackMagic II, Germany, Aixtron). A 2-nm layer of iron was
Ac ce p
deposited onto a Si substrate as a precursor, and then CNTs were grown on this substrate in an acetylene atmosphere via thermal CVD. The resulting CNTs are ~300 m in length, 6 nm in diameter, and approximately three carbon walls thick, as previously shown (Iijima et al, 2011). A section of the grown CNTs near the edge of the substrate was taken and drawn in parallel to the substrate surface while spinning. For the CNT yarn fabrication, the drawing speed and spinning rate were 40 mm/min and 4000 rpm, respectively. The fabricated CNT yarn was rolled on a paper rod (Fig. 1), and its structure was then observed via in situ TEM analysis.
Page 4 of 19
paper rod
an
Figure 1: Photograph of the drawn CNT yarns
us
3 mm
cr
ip t
CNT yarn
M
During current flow, the CNT yarn was observed and recorded by a digital camera
d
system (SC200D, Gatan, California, USA). A high-voltage TEM, equipped with a
te
digital camera system for in situ observation, is used for visual measurements (HVTEM,
Ac ce p
JEM-1000K RS, Akishima, Japan, JEOL). The CNT yarn was connected between two of the four electrodes equipped on the TEM holder. Typically, samples connected between these electrodes are fixed with screws. However, the CNT yarn diameter is too small to be fixed with a screw; and therefore, it was connected to the electrodes with a conducting paste (DOTITE D-550, Fujikura Kasei, Tokyo, Japan). After insertion into the TEM chamber, current
passed from the electrode through an electrical wire from
the TEM holder to outside of the chamber. A variable resistor controlled the current from a battery,
the electrodes carried the current to the CNT yarn. An electron energy
Page 5 of 19
loss spectroscope (EELS: Quantum, California, USA, Gatan) was used for elemental
current
ip t
analysis. Raman spectra were acquired for CNT yarns with and without exposure to to identify their crystallization. The Raman spectrometer was equipped with a
cr
532-nm wavelength laser (NRS-1000VUV, Tokyo, Japan, Jasco).
us
From TEM observations, the CNT yarn has a diameter of 5 m and some individual
an
CNTs protrude from the surface of yarn (Fig 2(a)). This suggests the yarn is not perfectly spun. Moreover, a closer examination of the edges of the yarn revealed spaces
M
between the CNTs (Fig. 2(b)). These spaces were further examined via electron energy
d
loss spectroscopy (EELS) (inset of Fig. 2(b)). The EELS spectra of the small dark
te
circular contrasts confirmed the presence of Fe-L3, L2, and O-K edges. Based on these
Ac ce p
results, it was determined that the circular contrasts are iron oxide particles. The ratio of the Fe-L3 and L2 intensity (L3/L2), which was measured from the deconvoluted Fe-L3 and L2 EELS spectra, gives the iron valence of iron oxide (Sparow et al., 1984). The L3/L2 ratio of iron oxide particle is ~3.3, suggesting the iron oxide particle is FeO. These FeO particles originate from the oxidized iron catalyst particles used for the CNT growth. Prior to the CNT growth, the iron catalyst is annealed after deposition by evaporation on the substrate surface. Our results show that the iron catalyst is expected to oxidize during the sample transfer and/or the fabrication of the CNT yarn.
Page 6 of 19
(b)
us
cr
ip t
(a)
200 nm
an
2 m
M
Figure 2: TEM images of (a) the center and (b) near the CNT yarn edge, inside the
d
dotted black line rectangle of Figure 2(a). The inset in Figure 2(b) shows the EELS
Ac ce p
te
acquired from inside the dotted line circle.
Results
Low-magnification TEM images of the CNT yarn over time are observed during a 5
mA current flow through the yarn containing FeO particles (Fig. 3). The centers of the TEM images contain thin, long contrast areas in which the black small catalyst particles are apparent. Only particles with diameters greater than ~250 nm are observed in the CNT yarn using low-magnification TEM During current flow, the large particles repeatedly move left and right on the yarn. In addition, both the imperfectly spun CNTs
Page 7 of 19
and the larger particles on the CNT yarn are observed to rotate around the center axis of
ip t
the yarn. Therefore, the CNT yarn is confirmed to rotate during current flow. This rotation phenomenon is not continuous; after one revolution, the rotation nearly stops.
cr
Further, when currents of 5 mA were applied to the yarn, rotation became negligible. As
us
a conductive past fixes both ends of the CNT yarn in place, the sole driving force for the
15s
30s
45s
te
75s
90s
Ac ce p
60s
d
M
0s
an
rotational phenomena is the current flow.
Figure 3: In situ TEM images of the CNT yarn over time during 5-mA current flow.
Next, the CNT yarn with a high current (26 mA) flow is observed through time
progressive TEM images (Fig. 4). The catalyst particles are illustrated in the small black circular area marked by white arrows. These catalyst particles decrease in size with time before completely disappearing. When current is applied to the yarn, its temperature
Page 8 of 19
rises past the melting point (MP) of FeO (1369 °C) via resistive heating (Robert and
ip t
Weast, 1973). Thus, the catalyst particles are expected to evaporate from the yarn. As the CNT yarn is not perfectly spun, the evaporated particles could escape from
Although the catalyst particles evaporated from the
cr
numerous pores within the yarn.
us
CNT yarn, its shape and diameter are unaffected. Due to similarities with graphene
significantly higher than the MP of FeO.
an
sheets, the MPs of the CNTs and graphene are approximately identical, which is Based on the evaporation of the catalysts, it
Ac ce p
te
d
M
is expected that current flow leads to the resistive heating of the CNT yarn.
Page 9 of 19
Figure 4: In situ TEM images of the CNT yarn over time during 26-mA current flow.
ip t
When a 31 mA current is applied to the CNT yarn, the CNT yarn disappears from observation. The TEM images show that the CNT yarn is cut from the holder (Fig. 5).
cr
The diameter of the cut end of the yarn is smaller than the diameter of the remaining
Furthermore, regions of the cut CNT yarn exhibit some amorphous carbon,
an
5(b)).
us
CNT yarn (Fig. 5(a)). The cut end is very small, consisting of only a few CNTs (Fig.
indicating that the CNT yarn sublimed from the resistive heating due to the large current
M
flow. Since the heat capacity of the electrodes is considerably higher than the yarn, the
d
CNT yarn heated considerably during the current flow, yet the temperature close to the
te
electrodes did not rise. The center of the CNT yarn between the electrodes heated to the
Ac ce p
highest temperature. Thus, the CNT sublimed in this region from the resistive heating and deposited on the remaining CNT yarn that was at a lower temperature. Furthermore, the diameter of the CNT yarn gradually decreases approaching the broken end of the yarn, due to the current flowing through the surface.
For the region of yarn with the
greatest degree of heating, there is a large surface area with a low current density prior to CNT sublimation through resistive heating. As the CNT yarn is heated, the CNT sublimation occurs over a large area with a high current flow. As the CNTs sublimates on the surface of the yarn, the surface area decreases and the current density increases.
Page 10 of 19
The increase in current density leads to further CNT sublimation around the center of
diameter of the cut end of the CNT yarn is expected to decrease.
(b)
M
an
us
cr
(a)
ip t
the reduced surface area. As the CNT sublimation increases at the highest heat, the
5 m
te
d
50 nm
Ac ce p
Figure 5: In situ TEM images of the cut CNT yarn after 31-mA current flow.
The Raman spectrum is acquired for the initial CNT yarn prior to current flow and the
cut CNT yarn (Fig. 6). Two peaks appear in each spectrum, with the D-band peak at ~1350 cm−1 and the G-Band peak at ~1580 cm−1. The G-band peak results from vibrations on the 6-membered ring of carbon atoms along the nanotube axis and circumference (Dresselhaus et al., 2005). The D-band peak results from amorphous or disordered carbon and defects in the CNT samples (Datshuk et al., 2008). Typically, the
Page 11 of 19
intensity ratio of the G-band to the D-band peaks (IG/ID) is used to determine the CNT
ip t
purity and crystallization. CNTs with a high IG/ID ration are purer with a higher degree of crystallinity than those with a low IG/ID ratio. The IG/ID ratio of the initial CNT yarn
cr
is 0.90, while that of the cut yarn is 1.52 (Fig. 6) As the CNT yarn temperature increases
us
due to resistive heating during current flow, the crystallization of individual CNTs
an
within the yarn improves due to high-temperature exposure, as no amorphous carbon is observed in the initial CNT yarn (Fig. 2). In general, the CNT crystallization improves
M
with heating over 2000 ºC under vacuum (Ci et al., 2001, Huang et al., 2004). The
d
progression of crystallization is not apparent from our observations, but the effect of the
Ac ce p
te
raised temperature on the CNT yarn is apparent in the Raman spectra.
Page 12 of 19
Figure 6: Raman spectra acquired from the CNT yarn before (initial) and after current
ip t
flow (of the cut CNT yarn).
cr
Discussion
us
The above results demonstrate an important phenomena occurring with the CNT yarn
an
during current flow. As the current flows through the CNT yarn, its temperature rises, and the behavior the yarn changes with increasing temperature. Initially, a rotation of
M
the CNT yarn is observed. This rotation phenomenon is expected based on the yarn’s
d
thermal expansion. The CNT yarn elongated with heating, and it is expected that the
te
yarn would expand along its diameter. However, with the yarn length in millimeters and
Ac ce p
diameter in micrometers, the thermal expansion coefficients along its length and diameter directions are nearly equal. Since the elongation of the CNT yarn is more apparent than its expansion, the thermal elongation must be considered when discussing the rotation phenomenon. The twist density of the CNT yarn decreases with elongation through thermal expansion. Therefore, the rotation phenomena results from twist unraveling. For the length of the CNT, the thermal expansion coefficient increases with temperature, becoming nearly saturated at ~1300 ºC (Jiang et al., 2004). Above 1300 ºC, the coefficient is essentially constant, ceasing the rotation phenomena. While nearly all
Page 13 of 19
materials elongate with heating, this rotation is distinctive of CNT yarns. This rotation
ip t
phenomenon of the CNT yarn may affect its application in electric circuits. If the temperature of the CNT yarn is not controlled when used as an electric wire, it will
cr
elongate and rotate. This could lead to the CNT yarn moving and interfering with the
us
surrounding electrical elements, which may result in the failure of the electrical circuit.
drawn CNT yarn may not be continuous.
an
The spinning of the yarn can be removed from the fabrication process, although the
M
The catalyst evaporation phenomena observed after the rotation is an important
d
property of CNT yarn. Previously, metal catalysts were removed by heating the CNT
te
under vacuum or exposing it to an acid solution. These purified CNTs were then drawn
Ac ce p
to obtain a CNT yarn without the metal catalyst. However, our results reveal that it is possible to obtain a CNT yarn without metal catalysts by applying a current during the drawing process. This would allow for electrical circuit fabrications with a CNT yarn. Finally, at large current flows, it is shown that the CNT yarn sublimates, breaking the
yarn. The sublimated carbon is deposited on the nearby, cooler areas of yarn, and it is amorphous with low resistivity. If incorporated, this could affect the properties of an electronic circuit if the CNT yarn (Miyagawa et al., 2006). However, the carbon
Page 14 of 19
sublimation temperature is at ~3550 ºC (Robert and Weast, 1973). Thus, the CNT yarn
ip t
is suitable for use as an electric wire in high-temperature electronic devices.
cr
Conclusions
us
During the application of different current levels, in situ TEM observations of CNT
an
yarns with oxidized catalyst particles revealed yarn rotation, catalyst evaporation, and a cutting phenomenon. It is believe that these phenomena result from resistive heating
M
during the current flow through the yarn. Moreover, it is clear that CNT yarns could be
d
used as electrical wires at high temperature, around carbon sublimation temperature,
te
and under vacuum. These phenomena need to be recognized when incorporating CNT
Ac ce p
yarns into novel electronic devices, particularly as wires. These phenomena could both positively and adversely affect the device.
Acknowledgements
This work was supported by the Advanced Characterization Nanotechnology Platform
of the National Institute for Materials Science and the High Voltage Electron Microscopy Laboratory of Nagoya University.
Page 15 of 19
References
ip t
Iijima S., 1991. Helical microtubules of graphitic carbon. Nature. 354, 56-58. Bachtold, A., Strunk, C., Salvetat, P.J., Bonard, M.J., Forró, L., Nussbaumer, T.,
cr
Schönenberger, C., 1999. Aharonov-Bohm oscillations in carbon nanotubes. Nature.
us
397, 673-675.
an
Zou, P.X., Abe, H., Shimizu, T., Ando, A., Nakayama, Y., Tokumoto, H., Zhu, M.S., Zhout, S.H., 2004. Simple thermal chemical vapor deposition synthesis and
M
electrical property of multi-walled carbon nanotubes. Physica E. 24, 14-18.
d
Wong, W.E., Sheehan, E.P., Liber, M.C., 1997. Nanobeam Mechanics. Elasticity,
te
Strength, and Toughness of Nanorods and Nanotubes. Science. 277, 1971-1975.
Ac ce p
Meo, M., Rossi, M., 2006. Tensile failure prediction of single wall carbon nanotube. Eng. Fract. Mechanics. 73, 2589-2599.
Jiang, K., Li, Q., Fan, S., 2002. Spinning continuous carbon nanotube yarns. Nature. 419, 801.
Zhang, M., Atkinson, R.K., Baughman, H.R., 2004. Multifunctional Carbon Nanotube Yarns by Downsizing an Ancient Technology. Science. 306, 1358-1361.
Page 16 of 19
Tran, D.C., Humphries, W., Smith, M.S., Huynh, C., Lucas, S., 2009. Improving the
ip t
tensile strenghth of carbon nanotube spun yarns using a modified spinning process. Carbon. 47, 2662-2670.
cr
Miao, M., 2011. Electrical conductivity of pure carbon nanotube yarns. Carbon. 49,
us
3755-3761.
an
Iijima, T., Oshima, H., Hayashi. Y., Suryavanshi, B.U., Hayashi, A., Tanemura, M., 2011. Morphology control of a rapidly grown vertically aligned carbon-nanotube
M
forest for yarn spinning. Phys. Stat. Sol. A. 10, 2332-2334.
te
CRC PRESS, Ohio
d
Robert, C., Weast, P.D., 1973. CRC Handbook of chemistry and physics 54th edition.
Ac ce p
Sparow T.G., Williams B.G., Rao C.N.R., Thomas J.M., 1984. L3/L2 WHITE-LINE INTENSITY RATIOS IN THE ELECTRON ENERGY-LOSS SPECTRA OF 3d TRANSITION-METAL OXIDES. Chem. Phys. Lett. 108, 547-550.
Dresselhaus, S.M., Dresselhaus, G., Saito, R., Jolio, A., 2005. Raman spectroscopy of carbon nanotubes. Phys. Rep. 409, 47-99.
Datshuk, V., Kalyva, M., Papagelis, K.P., Tasis, D., Siokou, A., Kallitsis, I., Galiotis, C., 2008. Chemical oxidation of multiwalled carbon nanotubes. Carbon. 46, 833-840.
Page 17 of 19
Ci, L., Wei, B., Xu, C., Liang, J., Wu, D., Xie, S., Zhou, W., Li, Y., Liu, Z., Tang, D.,
ip t
2001. Crystallizationh behavior of the amorphous carbon nanotubes prepared by the CVD method. J. Crystal Growth. 233, 823-828.
cr
Huang, W., Luo, G., Wei, F., 2003. 99.9% purity multi-walled carbon nanotubes by
us
vacuum high-temperature annealing. Carbon. 41, 2585-2590.
an
Jinag, H., Liu, B., Huang, Y., Hwang, C.K., 2004. Thermal Expansion of Single Wall Carbon Nanotubes. J. Eng. Mater. Tech. 126, 265-270.
M
Iijima, T., Oshima. H., Hayashi, Y., Suryavanshi, B.U., Hayashi, A., Tanemura, M.,
d
2012. In-situ observation of carbon nanotube yarn from vertically aligned carbon
te
nanotube forest. Diam. Rel. Mater. 24, 158-160.
Ac ce p
Miyagawa, S., Nakao, S., Choi, J., Ikeyama, M., Miyagawa, Y., 2006. Effects of target bias voltage on the electrical conductivity of DLC films deposited by PBII/D with a bipolar pulse. Nucl Inst. Meth. in Phys. Res. B. 242, 346-348.
Robert, C., Weast, P.D., 1973. CRC Handbook of chemistry and physics 54th edition. CRC PRESS, Ohio.
Page 18 of 19
Highlights
ip t
The CNT yarn during current flow was observed by TEM. After starting current flow, rotation and cutting phenomena on the CNT yarn were
cr
confirmed.
us
Moreover, catalyst, which was used for CNT growth, was evaporated.
Ac ce p
te
d
M
an
These phenomena occur due to resistive heating during current flow.
Page 19 of 19
