# Fileset

[Prb153404.pdf](https://mdr.nims.go.jp/filesets/3115002b-ddc4-4301-97ab-0882adb6249b/download)

## Creator

Toshiya Okazaki, Shunji Bandow, Goshu Tamura, Yoko Fujita, Konstantin Iakoubovskii, Said Kazaoui, Nobutsugu Minami, Takeshi Saito, Kazu Suenaga, Sumio Iijima

## Rights



## Other metadata

[Photoluminescence quenching in peapod-derived double-walled carbon nanotubes](https://mdr.nims.go.jp/datasets/edd4198e-cd1e-46fe-afe1-b188b7d9729b)

## Fulltext

Photoluminescence quenching in peapod-derived double-walled carbon nanotubesToshiya Okazaki,1,* Shunji Bandow,2 Goshu Tamura,2 Yoko Fujita,2 Konstantin Iakoubovskii,3 Said Kazaoui,1,3Nobutsugu Minami,3 Takeshi Saito,1 Kazu Suenaga,1 and Sumio Iijima1,21Research Center for Advanced Carbon Materials, National Institute of Advanced Industrial Science and Technology (AIST),Tsukuba 305-8565, Japan2Department of Materials Science and Engineering, 21st century COE (Nanofactory), Meijo University, 1-501 Shiogamaguchi,Tenpaku-ku, Nagoya 468-8502, Japan3Nanotechnology Research Institute, AIST, Tsukuba 305-8565, Japan�Received 22 July 2006; published 5 October 2006�Photoluminescence �PL� behavior of peapod-derived double-walled carbon nanotubes �DWNTs�—the sim-plest form among multiwalled carbon nanotubes—is investigated. Even though the optical absorption and theresonant Raman spectra show the characteristic features of DWNTs, the PL signals originated from DWNTsare severely suppressed. This suppression is a consequence of an interlayer interaction between the inner andthe outer tubes that efficiently quenches the PL signals of the DWNTs.DOI: 10.1103/PhysRevB.74.153404 PACS number�s�: 78.67.Ch, 73.22.�f, 78.30.NaAn important breakthrough in the carbon nanotube spec-troscopy came in 2002 when photoluminescence �PL� signalswere recorded from isolated single-walled carbon nanotubes�SWNTs� that were wrapped with sodium dodecyl sulfate�SDS�.1,2 Since then, two-dimensional PL excitation andemission mapping has been widely recognized as a powerfultool for the characterization of the unique electronic proper-ties of the SWNTs associated with their low dimensionality,and for the analysis of the detailed tubule distribution of thebulk samples.3 Although PL properties of SWNTs are wellestablished, it is yet unclear whether PL can be obtainedfrom MWNTs. In particular, one might expect that the innercore tube is protected from the environment and its intrinsicoptical properties are preserved.Double-walled carbon nanotubes �DWNTs� may answerto the above question because they have the simplest struc-ture among multiwalled carbon nanotubes �MWNTs�, whichmake it easier to interpret the experimental observations. Re-cently, PL signals have been detected from DWNTs synthe-sized by chemical vapor deposition �CVD� and attributed tothe inner shells of DWNTs.4 However, it should be noted thatCVD-grown DWNT samples usually are mixed with smallamounts of SWNTs that have similar diameters as the innercores of the DWNTs. Therefore, it is uncertain whether thedetected PL originated from the isolated SWNTs or trulyfrom the inner shells of DWNTs. Alternatively, DWNTs canbe synthesized by thermal treatment of SWNTs encapsulat-ing C60 molecules �the so-called “peapods”�.5 The peapod-derived DWNT samples also contain undesirable SWNTs,however there is a clear difference between the SWNTs ex-isting in the CVD sample and those of peapod-derivedDWNTs: The diameters of the residual SWNTs in the formersamples are similar to those of the inner DWNT shells and inthe latter samples to those of the outer shells. Consequently,the peapod-derived DWNTs are more suitable for studyingPL from the inner tubes. Furthermore, comparative measure-ments between DWNTs and the pristine SWNTs are avail-able for the peapod-derived samples. This allows us to inves-tigate the PL behavior of the outer tubes after forming theinner tube structures.Here we report a detailed study of PL from the peapod-derived DWNTs, together with optical absorption and reso-nant Raman spectroscopy. Our results reveal that the emis-sion from both the inner tubes and the outer ones iseffectively quenched due to their interlayer interaction.The DWNTs studied here were synthesized by annealingof C60 peapods at 1200 °C.5 The pristine SWNTs with diam-eter �dt� ranging in �1.2–1.4 nm were produced by thepulsed laser vaporization of an Fe-Ni–containing carbon tar-get. Aqueous micellar solutions of DWNTs were prepared ina similar way to the procedure described by Bachilo et al.2Briefly, DWNTs ��0.5 mg� were dispersed in �15 ml ofD2O containing 1 wt % of sodium dodecylbenzene sulfonate�SDBS� by sonication for 10 min using a 200 W homog-enizer �Sonics VCX500�.6,7 The resultant solution was thencentrifuged at 127 600 g for 1 h �Hitachi CP 100MX centri-fuge� and the supernatant of the upper �2/3 volume wascollected. Transmission electron microscope �TEM� imageswere taken with a JEOL 2100F microscope operated at120 kV. The TEM specimen was prepared by casting a fewdroplets of a DWNT-SDBS-D2O solution onto a carbon mi-crogrid. Optical absorption spectra were recorded with a Shi-madzu UV-3150 spectrometer. Low-resolution resonant Ra-man spectra were measured with a single-gratingmonochromator equipped with an InGaAs diode array and anappropriate rejection filter, using a tunable Ti-sapphire laser�Spectra Physics 3900S� for excitation. High-resolution�1 cm−1� Raman spectra were measured with a BrukerRFS100 Fourier-transform �FT� Raman spectrometer using1064 nm laser excitation. PL mapping in a detection range of950–1630 nm was carried out with a Horiba SPEX Fluo-rolog 3-2 Triax spectrometer equipped with a near-infraredphotomultiplier �Hamamatsu H9170-75�. PL mapping at alonger wavelength �1200–1800 nm� was performed with ahome-built setup utilizing a tunable Ti-sapphire laser for ex-citation and a Jasco FTIR-800 spectrometer, equipped withan IR-enhanced InGaAs diode for detection.Figure 1 shows the TEM images of SDBS-suspendedDWNTs after centrifugation. It is confirmed that the nano-tubes are largely and uniformly covered by the surfactantthat appears with typical image contrast for amorphous ma-terials. Although the isolated DWNTs were frequently ob-served, most of them formed bundles. We suspect that thePHYSICAL REVIEW B 74, 153404 �2006�1098-0121/2006/74�15�/153404�4� ©2006 The American Physical Society153404-1http://dx.doi.org/10.1103/PhysRevB.74.153404bundling took place when the sample was dried on the TEMgrid. The HRTEM images indicate that about one half of thenanotubes have the double-walled structure.Figure 2�a� presents an UV-vis-NIR absorption spectrumof a DWNT-SDBS-D2O solution, together with the referencespectrum of the pristine SWNT SDBS D2O. Those spectrashow characteristic peaks associated with the interband tran-sitions. The broad features in the 800–1100 nm range belongto the second van Hove transitions �E22S � in semiconductingnanotubes with dt�1.2–1.4 nm. The peaks in the ranges400–600 nm and 600–800 nm can be assigned to the thirdvan Hove transitions �E33S � of the semiconducting nanotubesand E11M transitions in the metallic nanotubes, respectively.Besides those features, DWNTs show prominent peaks at�1037 and �598 nm �marked by arrows�, which, based ontheir transition energies, can be attributed to the E11S andE22S transitions of the inner tubes, respectively. Note that theE11S transitions of the inner tubes �dt�0.7 nm� coincidentlyoverlap with the E22S features of the outer tubes�dt�1.4 nm�.In order to investigate the origin of the absorption peak at�1037 nm, resonant Raman spectra were recorded using dif-ferent excitation wavelengths. The top spectrum of Fig. 2�b�was recorded under 975 nm excitation, which should reso-nantly excite the E22S transitions of either outer tubes ofDWNTs or SWNTs �see Fig. 2�a��. Indeed, the radial breath-ing mode �RBM� of vibration of the outer DWNT shells �orof the residual SWNTs� is clearly observed at �200 cm−1,which corresponds to dt�1.3 nm. Another RBM appears at�345 cm−1 when the excitation wavelength �1038 nm� isresonant with the E11S absorption of the inner tubes. The peakintensity decreased at the slightly longer excitation wave-length of 1064 nm, whereas the RBM of the outer tubesrelatively increased. This behavior matches the wavelengthdependence of the absorption spectrum of the DWNTs�Fig. 2�a��, especially of the 1037 nm peak associated withthe inner tubes, thus suggesting a similar origin to the1037 nm absorption peak and the 345 cm−1 RBM.Figure 2�c� shows a high-resolution Raman spectrum of aDWNT-SDBS-D2O solution measured at 1064 nm excita-tion. A set of strong and narrow lines observed in the spectralrange from 310 to 370 cm−1 is a characteristic feature of theinner tubes of peapod-derived DWNTs.8 Those peaks havebeen recently assigned to the �6,4� and �6,5� tubes.9 Strong�6,4� components indicate that substantial amounts of �6,4�tubes belong to the inner shells of our DWNTs. Note that theE11S and E22S transition wavelengths of the �6,4� SWNT wereestimated at 873 and 578 nm, respectively, in the SDS mi-cellar solution.10 These wavelengths are shorter than thoseobserved in Fig. 2�a� ��1037 and �598 nm�. The redshiftsobserved in the present experiment are consistent with thetheoretical predictions11,12 and could be attributed to the in-teraction between the inner and outer tubes as well as the PLquenching, as discussed below.FIG. 1. �Color� HRTEM images of DWNT SDBS D2O. Scalebar=10 nm.FIG. 2. �a� UV-vis-NIR absorption spectra of DWNT-SDBS-D2O and SWNT-SDBS-D2O solutions. �b� Low-resolution resonantRaman spectra of a DWNT-SDBS-D2O solution excited at 975,1038, and 1064 nm. �c� High-resolution resonant Raman spectrumof a DWNT-SDBS-D2O solution excited at 1064 nm. �d� An ex-panded view of the high-resolution Raman spectrum of the innertubes.BRIEF REPORTS PHYSICAL REVIEW B 74, 153404 �2006�153404-2If the inner tubes were luminescent, then the PL associat-ing with the E11S transitions could be observed �at least� at�1037 nm by the E22S excitation at �598 nm. However, de-spite the high concentration of DWNTs, no PL could be de-tected under 598 nm excitation �see Fig. 3�. This implies thatPL from the inner tubes is severely quenched, possibly due tothe presence of the outer tubes. On the other hand, strong PLpeaks are observed in the excitation range of 800–950 nmand the detection range of 1400–1600 nm. Those peaks canbe unambiguously assigned to the E11S emission from thesemiconducting tubes with dt�1.2–1.4 nm due to photoex-citation of E22S transitions. One could expect that the excita-tion of the inner tubes would be transferred to the outerDWNTs resulting in additional PL from those outer tubes. Inthis case, extra PL signals should be observed at �598 nmexcitation. However, no such signals could be detected �seeFig. 3�.The absence of PL from the outer tubes under the photo-excitation of E22S transitions of the inner tubes implies thatthe efficient and nonradiative relaxation paths exist betweenthe van Hove states of E11 for the outer tubes. We then per-formed a detailed comparison of PL from the outer shells ofthe DWNTs and from the pristine SWNTs to check such arelaxation path. Figure 4�a� shows a 2D PL contour map of aDWNT-SDBS-D2O solution measured in the longer wave-length region. The map reveals an apparent decrease in PLsignals from the outer tubes with a larger diameter �e.g.,�12,7� and �10,9� tubes� as compared to the correspondingsignals from the pristine SWNTs �Fig. 4�b��. In order to dis-tinguish the difference between the DWNTs and SWNTs ofthe PL intensities, we plotted the PL intensity ratio ofDWNTs and the corresponding pristine SWNTs �IPLDW/ IPLSW� asa function of the tube diameter dt �Fig. 4�c�� Even though PLpeak features for the DWNTs and SWNTs are quite similar�Figs. 4�a� and 4�b��, the values of IPLDW/ IPLSW decrease step-wise at dt�1.27 nm, as indicated by the dotted line in Fig.4�c�. It is worth noticing that theoretical calculations predictthat dt�1.28 nm is the smallest limit of the tube diameter forencasing the C60 molecules,13 which is surprisingly close tothe threshold value for the PL reduction deduced in Fig. 4�c�.This agreement may suggest that the formation of the innershells in the peapod-derived DWNTs results in the quenchingof PL by the outer DWNTs.Simple geometrical consideration of the structural trans-formation from peapod to DWNT suggests the length of theinner tube to be �2/3 of the outer tube and its remainingFIG. 3. �Color� Two-dimensional �2D� photoluminescence con-tour map of a DWNT-SDBS-D2O solution.FIG. 4. �Color� 2D PL contour maps of �a� DWNT-SDBS-D2Oand �b� SWNT-SDBS-D2O solutions in the longer wavelength re-gion. �c� PL intensity ratio between DWNTs and SWNTs �IDW/ ISW�as a function of a tube diameter, where each PL intensity ratio wasnormalized to that of a �15,1� tube.BRIEF REPORTS PHYSICAL REVIEW B 74, 153404 �2006�153404-3space is empty even if the filling yield of C60 is 100%.5Assuming that DWNT formation results in the strong ��10times� quenching of PL from the outer tubes, we might, ex-pect a �70% reduction of PL from the DWNTs with dt�1.27 nm. However, our PL measurements indicate a de-crease of �30–60 % �Fig. 4�c��.14 This somewhat low-PLquenching efficiency might suggest that the filling yield ofC60 is not 100%, but �70%. Considering that the presentDWNT sample contains the residual SWNTs having dt�1.27 nm and thus are incapable of encapsulating C60 mol-ecules, this estimate is consistent with our TEM observationsof the C60 peapods ��60% �.The intertube interaction may cause a downshift of theenergy levels and even closure of the band gap. Such hybrid-ization of the electronic structures of the inner and outertubes has been argued theoretically.11,12 For example, the �*states of the inner �7,0� tube overlap with those of the outer�n ,0� tubes �n=16, 17, 19, and 20�, which causes the metal-lization of the resulting DWNTs.11Recently, a 1D metallic character for the inner tubes ofpeapod-derived DWNTs has been deduced from 13C-NMRstudies.15 Charge transfer between the inner and outerDWNT shells was suggested as one of the possible reasonsfor this metallic behavior. However, in case of such a chargetransfer, absorption and Raman signals from one of the shellsshould be significantly reduced. This is not observed in ourspectra, thus rendering the charge transfer unlikely.Another important prediction from the theory is that theinteraction between the inner and outer shells in the DWNTstrongly depends on the interlayer distance11,12—naturally,the larger the distance, the weaker the interaction. This ten-dency might explain the previous observation of PL fromCVD DWNTs.4 In the peapod-annealing technique, the innertubes are created secondarily by annealing the existingSWNTs, while the inner and outer shells are simultaneouslygrown in the CVD process. A recent resonant Raman studysuggests that the annealing first creates inner tubes with rela-tively narrow diameters, and then these diameters graduallyincrease during annealing, thereby decreasing the interlayerdistance.8 In the present case, the diameter of the outer tubeswas estimated to be 1.278–1.384 nm from the PL map �Fig.4�. The diameters of the possible inner �6,4� and �6,5� tubesare 0.692 and 0.757 nm, respectively.2 This indicates that theinterlayer distance is less than �0.346 nm. On the otherhand, CVD-DWNTs in Ref. 4 have a much broader diameterdistribution. The diameters were estimated to be 0.5–2.5 and1.3–3.3 nm for the inner and the outer tubes, respectively. Abroader diameter distribution might result in the broader dis-tribution of the interlayer spacing. It is therefore likely thatthe CVD sample contains a significant fraction of DWNTswith the larger interlayer distance �much more than0.346 nm�16 and they can show PL signals.It has been established that the PL behaviors of SWNTsare dominated by the Coulomb interaction between the opti-cally produced electron-hole pairs �exitons�.17 The excitonbinding energy was estimated to be �0.420 eV for 0.8 nmSWNTs. If the exciton effects are affected by forming adouble-wall structure, the binding energy is expected to de-crease due to the lack of the low dimensionality. Assumingthat the outer tubes observed here �Fig. 4� have the bindingenergy of �0.420 eV,18 the band-gap energies of the firstand second van Hove singularities become �1.3 and�1.8 eV, respectively, which corresponds to the emissionand excitation wavelengths of �1000 and �700 nm, respec-tively. Since the exciton effect might be smaller for DWNTs,PL peaks should be shifted to the shorter wavelength ranges1000–1400 nm for emission and 700–800 nm for excitationafter forming a double-wall structure. However, any addi-tional PL peak was not observed in the corresponding region�Fig. 3�. It is thus unlikely that the exciton effects on DWNTsare responsible for the PL quenching observed here.In summary, the experimental results suggest that theelectronic properties of DWNTs are rather complex andstrongly depend on their structures and synthesis conditions.The interlayer distance might be an important parameter thatgoverns the optical properties of DWNTs. A decrease in theinterlayer distance enhances the interaction between the in-ner and outer shells, resulting in the efficient PL quenching.Based on these results, one might anticipate low PL efficien-cies for MWNTs. Indeed, the interlayer distance in MWNTsis significantly smaller than in the DWNTs,19 thus resultingin strong PL quenching.We thank H. Kataura �AIST�, S. Okada �Tsukuba Univer-sity�, and T. Nakanishi �AIST� for fruitful discussions.*Author to whom correspondence should be addressed; Email ad-dress: toshi.okazaki@aist.go.jp1 M. J. O’Connell et al., Science 297, 593 �2002�.2 S. M. Bachilo et al., Science 298, 2361 �2002�.3 R. B. Weisman, in Applied Physics of Carbon Nanotubes: Fun-damentals of Theory, Optics and Transport Devices, edited by S.V. Rotkin and S. Subramoney �Springer, Berlin, 2005�, pp. 183–202.4 T. Hertel et al., Nano Lett. 5, 511 �2005�.5 S. Bandow et al., Chem. Phys. Lett. 337, 48 �2001�.6 T. Okazaki et al., Nano Lett. 5, 2618 �2005�.7 T. Okazaki et al., Chem. Phys. Lett. 420, 286 �2006�.8 S. Bandow et al., Chem. Phys. Lett. 384, 320 �2004�.9 R. Pfeiffer et al., Phys. Rev. B 72, 161404 �2005�.10 R. B. Weisman and S. M. Bachilo, Nano Lett. 3, 1235 �2003�.11 S. Okada and A. Oshiyama, Phys. Rev. Lett. 91, 216801 �2003�.12 W. Song et al., Chem. Phys. Lett. 414, 429 �2005�.13 S. Okada et al., Phys. Rev. Lett. 86, 3835 �2001�.14 Note that a comparison of absolute PL intensities between thedifferent samples includes several uncertain factors, such as con-centration. Here we assume that the PL intensities of �15,1�tubes are the same in the DWNT and SWNT samples, i.e., thatIPLDW/ IPLSW=1, because the �15,1� tube is too narrow to encapsulateC60 molecules �see text�.15 P. M. Singer et al., Phys. Rev. Lett. 95, 236403 �2005�.16 W. Ren et al., Chem. Phys. Lett. 359, 196 �2002�.17 F. Wang et al., Science 308, 838 �2005�.18 The binding energy of �0.420 eV is considered to be the upperlimit for the outer tubes �dt=1.278–1.384 nm� because it de-creases as the tube diameter increases �see T. Ando, J. Phys. Soc.Jpn. 74, 77 �2005��.19 T. Hiraoka et al., Chem. Phys. Lett. 382, 679 �2003�.BRIEF REPORTS PHYSICAL REVIEW B 74, 153404 �2006�153404-4