Dual Raman Features of Double Coaxial Carbon Nanotubes with N

The downshift of the G band is correlated with the decreases of electrical resistivity of carbon nanotubes regardless of N or B doping. View: PDF | PD...
0 downloads 7 Views 285KB Size
Download PDF
NANO LETTERS

Dual Raman Features of Double Coaxial Carbon Nanotubes with N-Doped and B-Doped Multiwalls

2005 Vol. 5, No. 12 2465-2469

Quan-Hong Yang,† Peng-Xiang Hou,† Masashi Unno,† Seigo Yamauchi,† Riichiro Saito,‡ and Takashi Kyotani*,† Institute of Multidisciplinary Research for AdVanced Materials, Tohoku UniVersity, Sendai 980-8577, Japan, and Department of Physics and CREST JST, Tohoku UniVersity, Aramaki, Aoba, Sendai 980-8578, Japan Received September 6, 2005; Revised Manuscript Received October 22, 2005

ABSTRACT Double coaxial carbon nanotubes with nitrogen (N)-doped and boron (B)-doped multiwalls possess composite Raman characteristics, originating not only from the outer N-doped but also from inner B-doped layers. Both N and B dopings result in substantial shifts of the characteristic D band and G band of sp2 carbon constituting nanotube walls but in different ways. The downshift of the G band is correlated with the decreases of electrical resistivity of carbon nanotubes regardless of N or B doping.

Carbon nanotubes (CNTs) offer the potential for further miniaturization of electronic devices as long as it is possible to synthesize them with defined structures.1,2 Selective doping of nitrogen (N) or boron (B) into carbon layers is regarded as one of the effective approaches to modify the electrical properties of CNTs and to constitute carbon-based nanosized electronics.3-8 We have first prepared double coaxial CNTs with outer N-doped and inner B-doped multiwalls (termed as NB-CNTs thereafter), in which N and B dopings change the intrinsic electrical properties of outer and inner carbon layers in different ways, respectively, by introducing free electrons and holes into host carbon π-conjugation systems.9 In other words, we built a junction within a nanotube through controllable doping in different carbon layers, and this kind of nanojunction possibly behaves as nanodevices.7-9 Raman spectroscopy provides a sensitive probe for electronic and phonon structures of carbon materials including doped ones.10,11 Researchers have confirmed that the doping-induced shifts of the Raman-active tangential G band result from the electronic structure changes (changes of Fermi level) in CNTs upon the doping by alkali metal or halogen.11 Also, a number of Raman studies have been made on N- or B-doped carbon materials including CNTs and carbon films. These doped materials usually possess defect-induced D band and tangential G-band as undoped sp2 carbon materials do,12-16 and in some cases, D and G bands have been found to be broadened and greatly overlapped due to the dopings.14,16 In * Corresponding author. E-mail: [email protected]. † Institute of Multidisciplinary Research for Advanced Materials, Tohoku University. ‡ Department of Physics and CREST JST, Tohoku University. 10.1021/nl051779j CCC: $30.25 Published on Web 11/18/2005

© 2005 American Chemical Society

this study, we performed a detailed comparative Raman study on NB-CNTs and some reference samples (undoped and solely N- or B-doped CNTs) to confirm if there exist Raman shifts due to N or B doping, since these dopings greatly modify the electronic structure. We, for the first time, have observed different characteristic shifts of Raman features corresponding to N or B doping, and these shifts are intimately related to the structural changes and the increase of electrical conductance of doped CNTs. Furthermore, this study has revealed that NB-CNTs possess a composite Raman feature, being derived not only from the N-doped but also from B-doped layers. As presented in ref 9, we employed a two-step chemical vapor deposition (CVD) (the first step of CH3CN CVD and the second step of C6H6 and BCl3 CVD) in the nanochannels of an anodic aluminum oxide (AAO) template to synthesize multiwalled NB-CNTs. This kind of nanotubes consists of four N-doped carbon layers and eight B-doped layers to constitute the outer and inner stacks, respectively. Using a similar template technique, we also prepared several reference CNTs: undoped CNTs (C-CNTs), N-doped single stack CNTs (N-CNTs), double coaxial CNTs with outer undoped and inner B-doped multiwalls (CB-CNTs), and double coaxial CNTs with outer undoped and inner N-doped multiwalls (CN-CNTs). The detailed synthesis procedure was described in refs 7-9. N-CNTs and CN-CNTs with different N content became available by adjusting the faction of nitrogen source (CH3CN) in the feedstock, and we had N-CNTs and CN-CNTs with N/C atomic ratios in the range of 0-0.1. Similarly, the B/C ratios of CB-CNTs were easily

controlled in the range of 0-0.06 by adjusting the fraction of boron source, BCl3, in the feedstock. As presented in ref 9, we have to note that single-stack B-doped CNTs cannot be prepared by the present technique possibly due to a direct reaction between BCl3 and AAO. The atomic ratios (N/C and B/C ratios) of all the samples were obtained by X-ray photoelectron spectroscopy (XPS) measurements with the error (0.05, and details are given elsewhere.8,9 Raman spectra were recorded using a micro-Raman spectroscopy (Seishin Trading Co.) equipped with a Spex 500M spectrometer and a liquid nitrogen cooled chargecoupled device detector (SpectrumOne, Instrument S.A.). All samples measured were excited with 488 nm (2.54 eV) and 514.5 nm (2.41 eV) lines emitted from an argon ion laser (GLG 3280S, NEC) and a 647.1 nm (1.92 eV) line emitted from a krypton ion laser (BeamLok 2065, Spectra-Physics Lasers Inc.). Rayleigh scattering was rejected with a notch filter. For all specimens, the Raman spectra were measured under the same conditions. Note that the Raman results presented in this study were obtained at the laser excitation energy of 2.54 eV, unless otherwise specified. All Raman lines obtained were treated with a linear baseline subtraction and normalized with respect to the value of the highest intensity and then deconvoluted to Lorentzian lines. In the curve-fitting processes, all parameters, i.e., peak position, height and width of all peaks, were adjusted to attain the smallest value of chi square (χ2) in the chi-square test. Here, the chi-square test is a statistic method to evaluate the coincidence between the fitted lines and experimental Raman lines, and the lower the value of χ2, the better the curvefitting. Since NB-CNTs contain coaxial N-doped and B-doped layers, we first discussed the Raman spectra of N-CNTs and CB-CNTs to identify different Raman features of N-doped and B-doped layers. Figure 1 shows a stack plot of normalized and curve-fitted Raman spectra of N-CNTs with different N/C atomic ratios, together with the reference undoped C-CNTs. Like most sp2 carbon materials, all these spectra in this study are characterized by a two-peak feature but with shifted phonon frequencies and various bandwidths. Undoped C-CNTs possess an overlapped D band (around 1372 cm-1) and G band (around 1597 cm-1) through Lorentzian curve-fitting, and the intensity ratio of D band/G band (ID/IG) is 1.96. In the template synthesis of C-CNTs, no metal catalyst was used and the CVD temperature was relatively low, and as a result, the obtained nanotubes are less crystallized and more defective than the usual ones prepared by the catalytic CVD and arc discharge methods.17,18 Compared with C-CNTs, N-CNTs have more overlapped Raman line shapes, also being fitted into two features, but these fitted features are accompanied with gradually broadened bandwidths and shifted peak frequencies with the increase of N fraction in N-CNTs. The values of ID/IG increase from 2.10 to 2.65 corresponding to the increase in the N/C ratios from 0.03 to 0.10, indicating that N doping is very effective in introducing defects into the structure of CNTs, the results being consistent with a large body of literature including our previous work.8,19-21 It is manifested 2466

Figure 1. Normalized and curve-fitted Raman spectra (laser energy, 2.54 eV) of N-CNTs with different N fractions and reference undoped C-CNTs.

that D-band upshifts while G-band downshifts with the introduction of N into carbon layers and the shifts are enhanced with the increase in the N/C atomic ratios of N-CNTs. Parts a and b of Figure 2 quantify the correlations between N fractions and the shifts of the D band and G band, respectively (the shifts of Raman features of NB-CNTs will be discussed later). On the basis of Figure 2a, upon N doping, the D band upshifts from 1372 to 1385 cm-1 corresponding to a very small N/C ratio change from 0 to 0.03, indicating that the N doping at the doping level as low as 0.03 may initiate apparent modifications to the fine structure, leading to the substantial upshift of D band. With further increase in the N/C ratios, the D band turns to a smooth upshift (upshifts from 13 to 19 cm-1 as the N/C ratio changes from 0.03 to 0.10) and the upshifts in this range are roughly proportional to the N/C atomic ratios. The bandwidth (full width at half-maximum) of the D band has a big increase (from 218 to 252 cm-1) with the N/C ratios from 0 to 0.03, Nano Lett., Vol. 5, No. 12, 2005

Figure 2. Dependence of frequency shifts of the D band (a) and G band (b) in Raman spectra (laser energy, 2.54 eV) upon the N fractions for N-doped CNTs. C-CNTs and NB-CNTs are specially denoted, and other circles belong to N-CNTs.

while the width only has a very small increase from 252 to 266 cm-1 when the N/C ratios change in the range from 0.03 to 0.10. These findings support the idea that the N doping at very low level introduces a significant modification to the fine structure while further increase in the doping level does not enhance this modification apparently. As shown in Figure 2b, the G band of N-CNTs shows reverse shifts as compared to the D band. The N dopings impart stepwise changes to the G band, and there exists a linear relationship between G-band downshifts and N/C ratios at all doping levels (N/ C: 0-0.10). Single-stack B-doped CNTs are difficult to prepare by using the present template technique, and thus we measured Raman spectra of CB-CNTs with different B/C ratios to probe the Raman features of B-doped carbon layers. It should be noted that CB-CNTs prepared here contain only two or three undoped carbon layers and many more B-doped layers (about 10-15 layers). Meanwhile, undoped carbon layers (i.e., C-CNTs) possess much lower intensity of the Raman spectrum than doped ones (N-CNTs and CB-CNTs) under the same recording conditions. Considering both factors, the undoped layers contribute little to the Raman spectra of our CB-CNTs, and it is reasonable to assume that the Raman features of CB-CNTs represent those of B-doped layers. As in the case of N-CNTs, with B-doping into carbon layers (see Figure 3), CB-CNTs are also characterized by broadened and shifted two-peak Raman features. When compared with the case of undoped C-CNTs, the value of ID/IG increases with the B doping into carbon layers (2.01 and 2.31 for the B/C ratios of 0.04 and 0.06, respectively), indicating that B doping also introduces defects to the carbon layers but to a smaller extent than the N doping, being consistent with our previous results.9 Both the D-band and the G-band downshift due to B doping, the latter possessing much pronounced shift than the former. As shown in Figure 4, with the increase of the B/C ratios in the B-doped layers, the shifts are enhanced for both cases (D band and G band). The G band has more apparent downshift from 0 to 21 cm-1 as the B/C ratios change from 0 to 0.06, while the D band has the downshift from 0 to 8 cm-1. As is the case with N doping (Figure 2), it is difficult to correlate shifts of the D band with the B Nano Lett., Vol. 5, No. 12, 2005

Figure 3. Normalized and curve-fitted Raman spectra (laser energy, 2.54 eV) of CB-CNTs with different B fractions and reference undoped C-CNTs.

fraction at all the B doping levels, but shifts of the G band are linearly associated with the B/C ratios for CB-CNTs in the whole range from 0 to 0.06. We also measured Raman spectra for N-CNTs (N/C: 0.100), CB-CNTs (B/C: 0.060), and C-CNTs at two other laser energies (1.92 and 2.41 eV). The phonon frequencies of D bands for both the doped and undoped samples changed with the laser excitation energies, and the dispersion of peak position roughly accords with the widely accepted correlation between the employed laser energy and the frequency change for D bands (53 cm-1/eV).10,11 It is also found that, at the different laser energies, G bands for all three samples have 2467

Figure 4. Dependence of frequency shift of the D band (a) and G band (b) in Raman spectra (laser energy, 2.54 eV) upon B fractions for B-doped CNTs.

Figure 5. Curve-fitted Raman spectra (laser energy, 2.54 eV) of NB-CNTs: blue lines, two-peak fitting and D band and G band; orange lines, four-peak fitting with D1, D2, G1, and G2 bands.

no substantial shifts. These experimental findings suggest that both D bands and G bands of the doped samples possess the same nature as those of the undoped C-CNTs. In addition, we found that the values of ID/IG for all cases (N-CNTs, CBCNTs, and C-CNTs) increase with the decrease of the employed laser energies and a linear realationship exists between the ID/IG and the laser energies. The substantial shifts of Raman features give solid evidence for the modifications of phonon and electron structures of carbon layers due to N or B doping. We next discuss the Raman spectrum of NB-CNTs with outer N-doped and inner B-doped layers (N/C atomic ratio for N-doped layers, 0.020; B/C for B-doped layers, 0.030). Figure 5 shows a normalized Raman spectrum of NB-CNTs. As expected, NB-CNTs have a slightly broader Raman line than the N-CNTs and CB-CNTs and the two-peak spectrum can also be deconvoluted to D bands (around 1377 cm-1, upshifted by 5 cm-1 compared with C-CNTs) and G bands (around 1588 cm-1, downshifted by 9 cm-1 compared with C-CNTs) with χ2 ) 0.00119 for this peak-fitting. The peak position of each band is between those of N-CNTs and CBCNTs, respectively, with a N/C ratio and a B/C ratio similar to those of N-doped layers and of B-doped layers of NB2468

CNTs. Accordingly, the shifts of the D band and G band of NB-CNTs do not accord with the linear relationship between the shifts of Raman features and N or B fraction as shown in Figure 2 or 4, in which N doping or B doping is solely involved. We built NB-CNTs with N-doped outer and B-doped inner layers, and hence the Raman feature should be a composite deriving not only from N-doped outer layers but also from B-doped inner layers. We therefore further deconvoluted the two-peak features to four Lorentzians: two sub-D bands (D1 and D2) and two sub-G bands (G1 and G2). Here, D1 and G1 are attributed to N-doped layers and D2 and G2 are from B-doped layers. When performing the curve-fitting, we assume that all these subbands do accord with the linear relationship between Raman frequencies and N or B fraction (Figures 2 and 4), and accordingly, the center of D1, D2, G1, and G2 bands are set as 1383, 1371, 1595, and 1586 cm-1, respectively. For example, on the basis of Figure 2, we can easily obtain the upshift of D1 (about 11 cm-1) for the N-doped layers of NB-CNTs according to a value at the N/C ratio (0.02) in the linear relationship. These four subbands also contribute to a good peak-fit (χ2 ) 0.00115) for the original Raman line, suggesting that NBCNTs possess a composite Raman feature of outer N-doped and inner B-doped layers. In brief, the N- and B-doping results in characteristic shifts of both the D band and G band, and with increase of the doping level, the shifts will be enhanced. The N-doping leads to the upshifts, whereas the B-doping leads to the downshifts of the D band, the former possessing much larger shifts than the latter. Both N- and B-doping contribute to the downshifts of the G-band, and the latter gives larger downshifts. It is well accepted that the D band may be activated or enhanced by in-plane substitutional heteroatoms or introduction of defects. According to the previous study,8,9 the N doping brings about new chemical structures such as quaternary nitrogen and pyridine into graphene layers while the B doping also leads to C-B bonding inlaid in the graphene sheets. Therefore, it is reasonable to argue that shift of the D band is attributed to the appearance of new types of disorders with the introduction of these new fine structures and, N doping Nano Lett., Vol. 5, No. 12, 2005

the shifts of the D band and G band for our doped CNTs cannot be explained solely in terms of atomic weight of the doping components, being higher or lower than that of carbon, since these shifts are not in line with different atomic weight of N and B at all. Further research is ongoing to understand the physics of the shifts of Raman features associated with N or B dopings, and this is a key point to explain the logarithmic dependence of resistivity upon the downshift of the G band. Acknowledgment. Dr. Q. H. Yang is grateful for financial support from the JSPS postdoctoral fellowship (P 03282) for foreign researchers. Figure 6. Correlation between electrical resistivity of CNTs8,9 and downshift of the G band in Raman spectra (laser energy, 2.54 eV).

and B doping result in different shifts due to the formation of the different fine structures. The G band represents the tangential mode vibrations of carbon atoms in graphene sheets, and shifts of the G-band are interpreted in terms of C-C expansion (or contraction) and the changes of electronic structure.11 For our doped CNTs, the N or B doping into graphene sheets brings about great modifications to fine structure as proved by XPS and X-ray diffractometer (XRD).8,9 It is well-known that N doping imparts free electrons and B doping introduces holes to graphene sheets, and both processes promote electron transfer between valance and conduction bands and greatly improved electric conductance. The shifts of the G band due to N and B dopings presented in this study are in accordance with the changes of electronic structure due to the dopings. It is likely that the B doping leads to more pronounced changes since this kind of doping is characterized by the larger shift than the N doping at a similar doping level. In the previous study,8,9 we have reported that both B and N dopings lead to the decrease of electrical resistivity of carbon layers and the B doping brings about a much greater decrease. When correlating these resistivity values (R, Ω‚ cm) to the shift (S, cm-1) of Raman features, as shown in Figure 6, regardless of the type of doping, we found that the shifts (S) of the G band of doped CNTs are linear with respect to the logarithmic decreases of resistivity (R) except for double coaxial NB-CNTs in the following equation: R ) 1150 exp(-0.27S). We also note that NB-CNTs have a much lower resistivity than that predicted by the logarithmic relationship. We will next identify the cause of the deviation and if there exists some kind of synergetic effect between the N-doped and the B-doped layers, which contributes to the enhanced drop of resistivity. We should also note that

Nano Lett., Vol. 5, No. 12, 2005

References (1) Saito, S. Science 1997, 278, 77. (2) Graham, A. P.; Duesberg, G. S.; Seidel, R. V.; Liebau, M.; Unger, E.; Pamler, W.; Kreupl, F.; Hoenlein, W. Small 2005, 1, 382. (3) Terrones, M. Int. Mater. ReV. 2004, 49, 325. (4) Terrones, M.; Grobert, N.; Terrones, H. Carbon 2002, 40, 1665. (5) Wei, B. Q.; Spolenak, R.; Kohler-Redlich, P.; Ruhle, M.; Arzt, E. Appl. Phys. Lett. 1999, 74, 3149. (6) Fuentes, G. G.; Borowiak-Palen, E.; Pichler, T.; Liu, X.; Graff, A.; Behr, G.; Kalenczuk, R. J.; Knupfer, M.; Fink, J. Phys. ReV. B 2003, 67, 035429. (7) Xu, W. H.; Kyotani, T.; Pradhan, B. K.; Nakajima, T.; Tomita, A. AdV. Mater. 2003, 15, 1087. (8) Yang, Q. H.; Xu, W. H.; Tomita, A.; Kyotani, T. Chem. Mater. 2005, 11, 2940. (9) Yang, Q. H.; Xu, W. H.; Tomita, A.; Kyotani, T. J. Am. Chem. Soc. 2005, 127, 8956. (10) Dresselhaus, M. S.; Dresselhaus, G.; Saito, R.; Jorio, A. Phys. Rep. 2005, 409, 47. (11) Dresselhaus, M. S.; Eklund, P. C. AdV. Phys. 2000, 49, 705. (12) Hsu, W. K.; Firth, S.; Redlich, P.; Terrones, M.; Terrones, H.; Zhu, Y. Q.; Grobert, N.; Schilder, A.; Clark, R. J. H.; Kroto, H. W.; Walton, D. R. M. J. Mater. Chem. 2000, 10, 1425. (13) Maultzsch, J.; Reich, S.; Thomsen, C.; Webster, S.; Czerw, R.; Carroll, D. L.; Vieira, S. M. C.; Birkett, P. R.; Rego, C. A. Appl. Phys. Lett. 2002, 81, 2647. (14) Roy, D.; Chhowalla, M.; Hellgren, N.; Clyne, T. W.; Amaratunga, G. A. J. Phys. ReV. B 2004, 70, 035406. (15) Webster, S.; Maultzsch, J.; Thomsen, C.; Liu, J.; Czerw, R.; Terrones, M.; Adar, F.; John, C.; Whitley, A.; Carroll, D. L. Mater. Res. Soc. Symp. Proc. 2003, M7.8.1. (16) Guerino, M.; Massi, M.; Maciel, H. S.; Otani, C.; Mansano, R. D.; Verdonck, P.; Libardi, J. Diamond Relat. Mater. 2004, 13, 316. (17) Iijima, S. Nature 1991, 354, 56. (18) Amelinckx, S.; Zhang, X. B.; Braerts, D.; Zhang, X. F.; Ivanov, V.; Nagy, J. B. Science 1994, 265, 635. (19) Terrones, M.; Redlich, P.; Grobert, N.; Trasobares, S.; Hsu, W. K.; Terrones, H.; Zhu, Y. Q.; Hare, J. P.; Reeves, C. L.; Cheetham, A. K.; Ru¨hle, M.; Kroto, H. W.; Walton, D. R. M. AdV. Mater. 1999, 11, 655. (20) Suenaga, K.; Yudasaka, M.; Colliex, C.; Iijima, S. Chem. Phys. Lett. 2000, 316, 365. (21) Sung, S. L.; Tsai, S. H.; Tseng, C. H.; Chiang, F. K.; Liu, X. W.; Shih, H. C. Appl. Phys. Lett. 1999, 74, 197.

NL051779J

2469