- Email: [email protected]
Raman studies of carbon nanotubes
Volume 202. number 6
CHEMICAL PHYSICS LETTERS
5 February 1993
Raman studies of carbon nanotubes H. Hiura, T.W. Ebbesen, K. Tanigaki Fundamental Research Laboratories,
NEC Corporation, 34 Miyukigaoka.
Tsukuba 305, Japan
and H. Takahashi Department of Chemistry, School ofScience and Engineering, Waseda University, Tokyo 169, Japan
Received 14 October 1992;in final form I 1 November 1992
First- and second-order Raman spectra of closed-carbon nanostructures (nanotubes and nanoparticles) have been measured and compared to those of glassy carbon and highly oriented pyrolitic graphite. It is shown that this novel material has unique Raman spectral features which should be useful in its identification. Furthermore, the results indicate that nanotubes possess a high degree of crystalline order.
1. Introduction
2. Experiment
The Raman scattering of carbon materials such as graphite and diamond has been analyzed in great detail [ 1-6 1. It offers a unique tool to characterize the materials of study since the amount of ordering and degree of sp2and sp3bonding leaves a unique Raman “fingerprint”. Furthermore, it has also been shown that Raman scattering depends on the size of the graphite crystals [ 1,2,6]. Fullerenes, such as Cc,,, have vibrational characteristics reflecting their molecular nature, that is, distinct from graphite and diamond [7,8]. Carbon nanotubes having the physical dimensions of both large fullerenes (several nanometers in diameter) and solid graphite (several micrometers in length) [ 91 should have their own vibrational properties. Having recently discovered how to produce carbon nanotubes in large quantities [ lo], we report here the results of the first- and second-order Raman scattering study which indeed shows that this novel material has unique vibrational features. The material contains both nanotubes and nanopat-ticles (closedshell graphitic structures of nanometer dimensions) [lo]; the results are discussed in terms of the presence of both these species.
Nanotubes were synthesized using the carbon arc plasma method, which has been described in detail elsewhere [ lo]. Briefly, a dc arc plasma is formed between two glassy graphitic rods in an inert atmosphere of helium. As the positive electrode is consumed, a deposit forms on the negative carbon electrode. This deposit is rod-shaped and composed of an inner soft black core and an outer hard metallic shell. Both transmittance electron microscope (TEM) and scanning electron microscope (SEM) measurements show that the inner core is filled with nanotubes and nanoparticles while the hard outer shell contains few of these. We measured the Raman spectra of both the inside and outside of the deposit with no further treatment. We also took Raman spectra of glassy carbon (density 1.7 g/cm3) and highly oriented pyrolitic graphite (HOPG) for comparison with the Raman spectra of nanotubes and nanoparticles. All Raman spectra were obtained using a polychromator (Spex, Triplemate 1877) equipped with a multichannel ICCD detector (Stanford Research). Strong scattering from the rough samples was sharply cut by use of a holographic edge filter (Physical Optics Corporation). Raman exci-
0009-2614/93/$
06.00 0 1993 Elsevier Science Publishers B.V. All rights resewed.
509
Volume202, number 6
5 February 1993
CHEMICAL PHYSICS LETTERS
tation was performed at room temperature by 532 nm laser light of a cw Nd: YAG laser (Coherent, Antares 76) with a typical incident laser power in the range of 200 to 300 mW over 20.3 mm2. Although the laser beam was polarized, the scattered light was collected at 90” without polarizers. The accumulation time was about 20 min. The spectral resolution was 4 cm-’ at the center of each spectrum. The Raman spectra between 400 and 1800 cm- ’were calibrated by the Raman spectrum of indene and those between 1800 and 3250 cm-’ by Ne lamp lines with an accuracy of ? 1 cm-‘. No smoothing was applied to the spectra.
3. Results and discussion Fig. 1 shows the Raman spectra of several carbon solids, (a) HOPG, (b) the inner core of the deposit containing nanotubes and nanoparticles, (c) the outer shell of deposit, (d) glassy carbon, the raw material from which nanotubes and nanoparticles are formed. In the HOPG spectrum, fig. la, a strong Raman band located at 1580 cm-’ is observed. HOPG crystal belongs to the space group D& and the irreducible representations of the vibrational modes arc expressed as F= 2E2,+ E, +A2, + 2Bzs [ 3,111. Among them only the two EZgmodes are Raman active. The strong Raman band at 1580 cm-’ corresponds to one of the EZgmodes, which has been assigned to the vibrational mode corresponding to the movement in opposite directions of two neighboring carbon atoms in a graphite sheet [ 2,12 1. The Raman spectrum of glassy carbon (fig. Id) is different from that of HOPG. The spectrum of glassy carbon exhibits a very strong Raman band located at 1348 cm-’ and a strong band at 1584 cm-’ with a shoulder at 1620 cm-‘. The band appearing around 1350 cm-’ in carbon solids is explained in terms of the relaxation of the wavevector selection rules resulting from finite crystal size effects, which allows the M point phonon to contribute to the Raman scattering [ 1,2]. The first-order Raman spectrum (fig. lb) of the inner core deposit, containing nanotubes and nanoparticles, shows a striking resemblance to that of HOPG with a very strong Raman band at 1574 cm-‘. This similarity means that structures of nan510
1800
1600
1400 1200 Y
1000
800
600
401)
J cm”
Fig. 1. Comparison of first-order Raman spectra of (a) highly oriented pyrolytic graphite (HOPG), (b) the inner core of the deposit containing nanotubes and nanoparticles, (c) the outer shell of the deposit, (d) glassy carbon.
otubes and nanoparticles resemble that of HOPG which has a perfect crystalline graphitic structure. However, HOPG is formed of two-dimensional flat sheets while the nanotubes are curved and closed graphitic structures. This probably accounts for the 6 cm-’ Raman band frequency shift with respect to that of HOPG. The band width around 1580 cm-’ of the nanotube mixture (23 cm-‘) is slightly wider than that of HOPG ( 15 cm-‘) but much narrower than the corresponding band of glassy carbon. This is surprising considering the large distribution of nanotubes and nanoparticles in size and length. The above results support the idea that carbon nanotubes are indeed nearly perfect crystalline structures [ 9 1. In other words the sample acts as a collection of micro-crystals of different sizes with few defects. Another significant difference between figs. 1a and lb is that an additional weak Raman band appears
Volume 202. number 6
CHEMICAL PHYSICS LETTERS
around 1350 cm-’ in the nanotube and nanoparticle sample. As discussed earlier, it has been well established that the appearance of this band can be ascribed to the presence of graphitic particles and crystallites of finite size of nanometer order. Furthermore, its intensity relative to the 1580 cm-’ band varies as the inverse of the crystal planar domain size [ 1,2,6]. From this result and the second-order Raman measurements described later, the 1349 cm-’ band is best assigned to the presence of nanoparticles in the sample which are several nanometers in size. Knight and White demonstrated a relationship between the Raman intensity of the 1350 cm-’ band with respect to that of the 1580 cm- ’band, and the crystal size of planar graphite [ 6 1, It is not clear whether this is directly applicable to nanoparticles since these are highly curved and closed structures where the zone-boundary conditions and consequently the wavevector selection rule relaxation might be different [ l-31. At first glance the hard outer shell of the deposit looks rather like glassy carbon (with a similar density). However, the Raman spectrum shown in fig. 1c resembles that of HOPG and specially that of the inner core deposit. Nevertheless, the band around 1580 cm-’ shifts to higher frequency with respect to that of HOPG and the band around 1350 cm-’ is more intense than that of the inner core. These,Raman observations indicate that the outer shell is formed from the fusion (sintering) of nanotubes and nanoparticles into a solid in the reaction chamber. These nanostructures are then inseparable from the solid matrix. The second-order Raman spectra of various forms of graphite have been studied by several researchers [ 2,5,6]. Fig. 2 shows the second-order Raman spectra from 2350 to 3300 cm-’ of the same carbon solids as in fig. 1. No Raman bands were observed in the range 1800-2350 cm- ’ for these carbon solids. The Raman spectra of HOPG and glassy carbon, figs. 2a and 2d, respectively, are in good agreement with those reported in the literature. Strong Raman bands around 2700 cm-’ and weak bands around 2460 and 3250 cm-’ are observed in HOPG and glassy carbon spectra. The dominating Raman band around 2700 cm-’ has been assigned to the overtone of the firstorder mode at x 1350 cm-’ [ 2,5]. In the glassy carbon spectra, an additional broad band around 2945
5 February 1993
(a)
Fig. 2. Comparison of the second-order Raman spectra of (a) HOPG, (b) the inner core of the deposit containing nanotubes and nanoparticles, (c) the outer shell of the deposit, (d) glassy carbon (these are the same samples as in fig. 1).
cm-’ is observed, which is thought to arise from a combination of the strong Raman modes at 1350 and 1620 cm-’ [ 21. The Raman spectra of the inner core of the deposit (fig. 2b) exhibits a relatively simple feature in this region; only two bands, a strong Raman band 2687 cm-’ and a weak band around 2460 cm-‘, can be recognized. The band located at 2687 cm-’ can be safely assigned to an overtone of the first-order mode around 1350 cm-‘. However this band shows a large downshift in frequency with respect to both the 2719 cm-’ band of HOPG and the 2696 cm-’ band of glassy carbon. Furthermore, the 2687 cm-’ band is narrower in width than the corresponding band of glassy carbon, and has no structure while the corresponding band of HOPG has a shoulder around 2684 cm-‘. These observations indicate that the 2687 cm-’ band is unique to nanotubes and nanoparticles. Another possibility is that the shoulder band of HOPG at 2684 cm-’ and the nanotube peak at 2687 cm-’ are related. The prob511
Volume 202, number 6
CHEMICAL PHYSICS LETTERS
lem is that this shoulder of the HOPG band has never been clearly assigned. One possibility is that it is related to the dimensionality of the material. Further investigation of this feature is necessary. The second-order Raman spectrum of the outer shell shows similarity to both the inner core material and glassy carbon (lig. 2~). As with the first-order Raman results, it supports the idea that this material is composed of sintered nanotubes and nanoparticles.
4. Conclusion The first- and second-order Raman spectra of closed-carbon nanotubes and nanoparticles were measured for the first time. Comparison of these spectra with those of HOPG and glassy carbon reveals that the nanotubes and nanoparticles have unique structures distinct from those of flat graphitic sheets of HOPG and possess a high degree of crystallinity. This also suggests that other physical properties such as mechanical strength and conductivity should reflect features of pure single crystals. Finally these results demonstrate that Raman spectroscopy is a powerful tool not only for identifying the presence of nanotubes and nanoparticles but also for studying their structures and properties.
512
5 February 1993
Acknowledgement The authors are grateful to P.M. Ajayan, N. Hamada and H. Igarashi for their useful comments with regards to this study.
References [ 1] F. Tuinstra and J.L. Koenig, J. Chem. Phys. 53 ( 1970) 1126. [2] R.J. Nemanich and S.A. Solin, Phys. Rev. I320 (1979) 392. [ 31 S.A. Solin, Physica 99B ( 1980) 443. [4] P. Lespade, R. Al-Jishi and M. S. Dressehause, Carbon 20 (1982) 427. [S] P. Lespade, A. Marchand, M. Couzi and F. Cruege, Carbon 22 (1984) 375. [6] D.S. Knight and W.B. White, J. Mater. Res. 4 (1984) 385. [7]D.S. Bethune, G. Meijer, W.C. Tang, H.J. Rosen, W.G. Golden, H. Seki, C.A. Brown and MS. de Vries, Chem. Phys. Letters 179 (1991) 181. [S] R.E. Stanton and M.D. Newton, J. Phys. Chem. 92 ( 1988) 2141. [9] S. Iijima, Nature 354 (1991)56. [IO] T.W. Ebbesen, P.M. Ajayan, Nature 358 (1992) 220. [ I1 ] K.K. Mani and R. Ramani, Phys. Stat. Sol. B 61 ( 1974) 659. [ 121R.J. Nemanich, G. Lucovsky and S.A. Solin, in: Proceedings of the International Conference on Lattice Dynamics, ed. M. Balkanski (Flammarion, Paris, 1978) p. 619.
CHEMICAL PHYSICS LETTERS
5 February 1993
Raman studies of carbon nanotubes H. Hiura, T.W. Ebbesen, K. Tanigaki Fundamental Research Laboratories,
NEC Corporation, 34 Miyukigaoka.
Tsukuba 305, Japan
and H. Takahashi Department of Chemistry, School ofScience and Engineering, Waseda University, Tokyo 169, Japan
Received 14 October 1992;in final form I 1 November 1992
First- and second-order Raman spectra of closed-carbon nanostructures (nanotubes and nanoparticles) have been measured and compared to those of glassy carbon and highly oriented pyrolitic graphite. It is shown that this novel material has unique Raman spectral features which should be useful in its identification. Furthermore, the results indicate that nanotubes possess a high degree of crystalline order.
1. Introduction
2. Experiment
The Raman scattering of carbon materials such as graphite and diamond has been analyzed in great detail [ 1-6 1. It offers a unique tool to characterize the materials of study since the amount of ordering and degree of sp2and sp3bonding leaves a unique Raman “fingerprint”. Furthermore, it has also been shown that Raman scattering depends on the size of the graphite crystals [ 1,2,6]. Fullerenes, such as Cc,,, have vibrational characteristics reflecting their molecular nature, that is, distinct from graphite and diamond [7,8]. Carbon nanotubes having the physical dimensions of both large fullerenes (several nanometers in diameter) and solid graphite (several micrometers in length) [ 91 should have their own vibrational properties. Having recently discovered how to produce carbon nanotubes in large quantities [ lo], we report here the results of the first- and second-order Raman scattering study which indeed shows that this novel material has unique vibrational features. The material contains both nanotubes and nanopat-ticles (closedshell graphitic structures of nanometer dimensions) [lo]; the results are discussed in terms of the presence of both these species.
Nanotubes were synthesized using the carbon arc plasma method, which has been described in detail elsewhere [ lo]. Briefly, a dc arc plasma is formed between two glassy graphitic rods in an inert atmosphere of helium. As the positive electrode is consumed, a deposit forms on the negative carbon electrode. This deposit is rod-shaped and composed of an inner soft black core and an outer hard metallic shell. Both transmittance electron microscope (TEM) and scanning electron microscope (SEM) measurements show that the inner core is filled with nanotubes and nanoparticles while the hard outer shell contains few of these. We measured the Raman spectra of both the inside and outside of the deposit with no further treatment. We also took Raman spectra of glassy carbon (density 1.7 g/cm3) and highly oriented pyrolitic graphite (HOPG) for comparison with the Raman spectra of nanotubes and nanoparticles. All Raman spectra were obtained using a polychromator (Spex, Triplemate 1877) equipped with a multichannel ICCD detector (Stanford Research). Strong scattering from the rough samples was sharply cut by use of a holographic edge filter (Physical Optics Corporation). Raman exci-
0009-2614/93/$
06.00 0 1993 Elsevier Science Publishers B.V. All rights resewed.
509
Volume202, number 6
5 February 1993
CHEMICAL PHYSICS LETTERS
tation was performed at room temperature by 532 nm laser light of a cw Nd: YAG laser (Coherent, Antares 76) with a typical incident laser power in the range of 200 to 300 mW over 20.3 mm2. Although the laser beam was polarized, the scattered light was collected at 90” without polarizers. The accumulation time was about 20 min. The spectral resolution was 4 cm-’ at the center of each spectrum. The Raman spectra between 400 and 1800 cm- ’were calibrated by the Raman spectrum of indene and those between 1800 and 3250 cm-’ by Ne lamp lines with an accuracy of ? 1 cm-‘. No smoothing was applied to the spectra.
3. Results and discussion Fig. 1 shows the Raman spectra of several carbon solids, (a) HOPG, (b) the inner core of the deposit containing nanotubes and nanoparticles, (c) the outer shell of deposit, (d) glassy carbon, the raw material from which nanotubes and nanoparticles are formed. In the HOPG spectrum, fig. la, a strong Raman band located at 1580 cm-’ is observed. HOPG crystal belongs to the space group D& and the irreducible representations of the vibrational modes arc expressed as F= 2E2,+ E, +A2, + 2Bzs [ 3,111. Among them only the two EZgmodes are Raman active. The strong Raman band at 1580 cm-’ corresponds to one of the EZgmodes, which has been assigned to the vibrational mode corresponding to the movement in opposite directions of two neighboring carbon atoms in a graphite sheet [ 2,12 1. The Raman spectrum of glassy carbon (fig. Id) is different from that of HOPG. The spectrum of glassy carbon exhibits a very strong Raman band located at 1348 cm-’ and a strong band at 1584 cm-’ with a shoulder at 1620 cm-‘. The band appearing around 1350 cm-’ in carbon solids is explained in terms of the relaxation of the wavevector selection rules resulting from finite crystal size effects, which allows the M point phonon to contribute to the Raman scattering [ 1,2]. The first-order Raman spectrum (fig. lb) of the inner core deposit, containing nanotubes and nanoparticles, shows a striking resemblance to that of HOPG with a very strong Raman band at 1574 cm-‘. This similarity means that structures of nan510
1800
1600
1400 1200 Y
1000
800
600
401)
J cm”
Fig. 1. Comparison of first-order Raman spectra of (a) highly oriented pyrolytic graphite (HOPG), (b) the inner core of the deposit containing nanotubes and nanoparticles, (c) the outer shell of the deposit, (d) glassy carbon.
otubes and nanoparticles resemble that of HOPG which has a perfect crystalline graphitic structure. However, HOPG is formed of two-dimensional flat sheets while the nanotubes are curved and closed graphitic structures. This probably accounts for the 6 cm-’ Raman band frequency shift with respect to that of HOPG. The band width around 1580 cm-’ of the nanotube mixture (23 cm-‘) is slightly wider than that of HOPG ( 15 cm-‘) but much narrower than the corresponding band of glassy carbon. This is surprising considering the large distribution of nanotubes and nanoparticles in size and length. The above results support the idea that carbon nanotubes are indeed nearly perfect crystalline structures [ 9 1. In other words the sample acts as a collection of micro-crystals of different sizes with few defects. Another significant difference between figs. 1a and lb is that an additional weak Raman band appears
Volume 202. number 6
CHEMICAL PHYSICS LETTERS
around 1350 cm-’ in the nanotube and nanoparticle sample. As discussed earlier, it has been well established that the appearance of this band can be ascribed to the presence of graphitic particles and crystallites of finite size of nanometer order. Furthermore, its intensity relative to the 1580 cm-’ band varies as the inverse of the crystal planar domain size [ 1,2,6]. From this result and the second-order Raman measurements described later, the 1349 cm-’ band is best assigned to the presence of nanoparticles in the sample which are several nanometers in size. Knight and White demonstrated a relationship between the Raman intensity of the 1350 cm-’ band with respect to that of the 1580 cm- ’band, and the crystal size of planar graphite [ 6 1, It is not clear whether this is directly applicable to nanoparticles since these are highly curved and closed structures where the zone-boundary conditions and consequently the wavevector selection rule relaxation might be different [ l-31. At first glance the hard outer shell of the deposit looks rather like glassy carbon (with a similar density). However, the Raman spectrum shown in fig. 1c resembles that of HOPG and specially that of the inner core deposit. Nevertheless, the band around 1580 cm-’ shifts to higher frequency with respect to that of HOPG and the band around 1350 cm-’ is more intense than that of the inner core. These,Raman observations indicate that the outer shell is formed from the fusion (sintering) of nanotubes and nanoparticles into a solid in the reaction chamber. These nanostructures are then inseparable from the solid matrix. The second-order Raman spectra of various forms of graphite have been studied by several researchers [ 2,5,6]. Fig. 2 shows the second-order Raman spectra from 2350 to 3300 cm-’ of the same carbon solids as in fig. 1. No Raman bands were observed in the range 1800-2350 cm- ’ for these carbon solids. The Raman spectra of HOPG and glassy carbon, figs. 2a and 2d, respectively, are in good agreement with those reported in the literature. Strong Raman bands around 2700 cm-’ and weak bands around 2460 and 3250 cm-’ are observed in HOPG and glassy carbon spectra. The dominating Raman band around 2700 cm-’ has been assigned to the overtone of the firstorder mode at x 1350 cm-’ [ 2,5]. In the glassy carbon spectra, an additional broad band around 2945
5 February 1993
(a)
Fig. 2. Comparison of the second-order Raman spectra of (a) HOPG, (b) the inner core of the deposit containing nanotubes and nanoparticles, (c) the outer shell of the deposit, (d) glassy carbon (these are the same samples as in fig. 1).
cm-’ is observed, which is thought to arise from a combination of the strong Raman modes at 1350 and 1620 cm-’ [ 21. The Raman spectra of the inner core of the deposit (fig. 2b) exhibits a relatively simple feature in this region; only two bands, a strong Raman band 2687 cm-’ and a weak band around 2460 cm-‘, can be recognized. The band located at 2687 cm-’ can be safely assigned to an overtone of the first-order mode around 1350 cm-‘. However this band shows a large downshift in frequency with respect to both the 2719 cm-’ band of HOPG and the 2696 cm-’ band of glassy carbon. Furthermore, the 2687 cm-’ band is narrower in width than the corresponding band of glassy carbon, and has no structure while the corresponding band of HOPG has a shoulder around 2684 cm-‘. These observations indicate that the 2687 cm-’ band is unique to nanotubes and nanoparticles. Another possibility is that the shoulder band of HOPG at 2684 cm-’ and the nanotube peak at 2687 cm-’ are related. The prob511
Volume 202, number 6
CHEMICAL PHYSICS LETTERS
lem is that this shoulder of the HOPG band has never been clearly assigned. One possibility is that it is related to the dimensionality of the material. Further investigation of this feature is necessary. The second-order Raman spectrum of the outer shell shows similarity to both the inner core material and glassy carbon (lig. 2~). As with the first-order Raman results, it supports the idea that this material is composed of sintered nanotubes and nanoparticles.
4. Conclusion The first- and second-order Raman spectra of closed-carbon nanotubes and nanoparticles were measured for the first time. Comparison of these spectra with those of HOPG and glassy carbon reveals that the nanotubes and nanoparticles have unique structures distinct from those of flat graphitic sheets of HOPG and possess a high degree of crystallinity. This also suggests that other physical properties such as mechanical strength and conductivity should reflect features of pure single crystals. Finally these results demonstrate that Raman spectroscopy is a powerful tool not only for identifying the presence of nanotubes and nanoparticles but also for studying their structures and properties.
512
5 February 1993
Acknowledgement The authors are grateful to P.M. Ajayan, N. Hamada and H. Igarashi for their useful comments with regards to this study.
References [ 1] F. Tuinstra and J.L. Koenig, J. Chem. Phys. 53 ( 1970) 1126. [2] R.J. Nemanich and S.A. Solin, Phys. Rev. I320 (1979) 392. [ 31 S.A. Solin, Physica 99B ( 1980) 443. [4] P. Lespade, R. Al-Jishi and M. S. Dressehause, Carbon 20 (1982) 427. [S] P. Lespade, A. Marchand, M. Couzi and F. Cruege, Carbon 22 (1984) 375. [6] D.S. Knight and W.B. White, J. Mater. Res. 4 (1984) 385. [7]D.S. Bethune, G. Meijer, W.C. Tang, H.J. Rosen, W.G. Golden, H. Seki, C.A. Brown and MS. de Vries, Chem. Phys. Letters 179 (1991) 181. [S] R.E. Stanton and M.D. Newton, J. Phys. Chem. 92 ( 1988) 2141. [9] S. Iijima, Nature 354 (1991)56. [IO] T.W. Ebbesen, P.M. Ajayan, Nature 358 (1992) 220. [ I1 ] K.K. Mani and R. Ramani, Phys. Stat. Sol. B 61 ( 1974) 659. [ 121R.J. Nemanich, G. Lucovsky and S.A. Solin, in: Proceedings of the International Conference on Lattice Dynamics, ed. M. Balkanski (Flammarion, Paris, 1978) p. 619.