J. Phys. Chem. C 2009, 113, 1751–1757 1751 Large Variety of Behaviors for the Raman G′ Mode of Single Walled Carbon Nanotubes upon Electrochemical Gating Arising from Different (n,m) of Individual Nanotubes Martin Kalbac,*,†,§ Ladislav Kavan,† Hootan Farhat,‡ Jing Kong,§ and Mildred S. Dresselhaus§,| J. HeyroVský Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, V.V.i., DolejškoVa 3, CZ-18223 Prague 8, Czech Republic, and Department of Materials Science and Engineering, Department of Electrical Engineering and Computer Science, and Department of Physics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 ReceiVed: October 5, 2008; ReVised Manuscript ReceiVed: NoVember 27, 2008 The dependence of the second-order Raman G′ mode of individual single walled carbon nanotubes (SWCNTs) as a function of electrochemical gating has been studied using in situ Raman spectroelectrochemistry. We show that the change of the frequency of the G′ mode with electrode potential (δ(ωG′)/δ(V)) is specific for different (n,m) of individual SWCNTs. The study of the G′ mode at the single nanotube level allowed us to observe many effects that get averaged in observations for the SWCNT bundles. These effects included an unusual dependence of the G′ mode frequency on electrode potential (V) for particular individual tubes, namely a decrease of the G′ mode frequency with increasing magnitude of electrode potential. This occurs for both increasing and decreasing V relative to V ) 0. It is demonstrated that there is no simple dependence of δ(ωG′)/δ(V) on tube diameter or excitation laser energy. Furthermore, the G′ mode intensity drops with an increase or decrease of the electrode potential relative to V ) 0 and this decrease in the G′ mode intensity was found to be gradual and not abrupt, which suggests a more complicated mechanism than the simple bleaching of electronic transitions. These observations give guidance to future theoretical work on this topic. Introduction Single walled carbon nanotubes (SWCNTs) have outstanding electronic properties with a large number of prospective applications in nanoscale electronics and devices such as displays, sensors, and supercapacitors. Resonance Raman spectroscopy is one of the most important methods for the study of SWCNTs. The main components of the Raman spectra of SWCNTs are the radial breathing mode (RBM), the tangential displacement band (TG), the disorder-induced mode (D), and the high-frequency second-order mode (G′). The tangential displacement band (also called the G band) is observed in the region of 1450-1600 cm-1, and in the case of metallic tubes, one of the peaks exhibits a pronounced Breit-Wigner-Fano (BWF) spectral line shape. The RBM frequency of SWCNTs is inversely proportional to the tube diameter. The D and G′ modes are observed in all kinds of polycrystalline carbon materials with sp2 hybridization. However, their physical origin has been explained only recently in terms of double-resonance theory.1-3 The D and G′ modes are observed in the spectral regions of 1250-1450 and 2500-2900 cm-1, respectively. The one-phonon second-order Raman D band appears only if there is a breakdown in translational crystal symmetry, which can be caused by defects in the structure. On the other hand, the twophonon second-order Raman G′ feature occurs independently of the presence of structural defects. The two phonons contributing to the G′ feature have wave vectors q and -q, and thus the momentum conservation constraint is automatically preserved. * Corresponding author. Telephone: 420 2 6605 3804. Fax: 420 2 8658 2307. E-mail: [email protected]. † Academy of Sciences of the Czech Republic. ‡ Department of Materials Science and Engineering, MIT. § Department of Electrical Engineering and Computer Science, MIT. | Department of Physics, MIT. Tuning and gaining an understanding of the electronic structure of SWCNTs are crucial for their application in nanoelectronic devices. Doping of SWCNTs can be carried out chemically or electrochemically. Electrochemical charging is more favorable for fundamental studies due to the precise and easy control of the doping level.4 This inspired several spectroelectrochemical studies on SWCNTs,5-7 double walled carbon nanotubes (DWCNTs),8 and fullerene peapods.9 However, these measurements have been typically performed on bundled samples. The use of bundled samples complicated the interpretation of the results since the individual properties of specific tubes were averaged and modified by bundling.10 Therefore, it is desirable to focus on individual carbon nanostructures in fundamental studies. Individual SWCNTs can be obtained by (i) sonication of SWCNT bundles in an aqueous solution of surfactants such as sodium dodecyl sulfate (SDS),11 (ii) detachment of the nanotubes from bundles using surface adhesion forces,12 or (iii) their direct growth by chemical vapor deposition (CVD).13 The dispersion of bundles in SDS provides a sample suitable for photoluminescence measurements,11 which proves that successful debundling has occurred. Nevertheless, it is very difficult to remove the surfactant quantitatively, and thus this method is not favored for spectroelectrochemical measurements. The detaching of individual SWCNTs from a bundle is a clean and easy technique, and can be applied also to other nanostructures such as peapods and DWCNTs.12 On the other hand, the debundling is not perfect in this case. Although it provides some isolated tubes, most of the sample consists of substrate-bonded nanoribbons of interacting nanotubes. Furthermore, the localization of nanotubes on a substrate is random in both of the abovementioned techniques, and thus contacting nanotubes by external electrical leads is difficult. The CVD approach, however, can 10.1021/jp808797c CCC: $40.75 2009 American Chemical Society Published on Web 01/13/2009 1752 J. Phys. Chem. C, Vol. 113, No. 5, 2009 Kalbac et al. Figure 1. (a) Scanning electron micrograph of as-grown SWCNTs on a SiO2/Si substrate. The arrow indicates the direction of nanotube growth. (b) AFM image of an as-grown individual SWCNT (the measured height of the nanotube is 1.23 nm, and the RMS background level is 0.3 nm). provide clean long isolated tubes with a defined orientation on a substrate.14 Therefore this method is the most favorable for preparing samples for spectroelectrochemical measurements. Here we present in situ Raman spectroelectrochemical data on the G′ mode of individual SWCNTs. We focused our study on the change of the frequency of the G′ mode during electrochemical charging of SWCNTs. For bundled SWCNTs, depending on the nanotube sample, the laser excitation energy, and the conditions of electrochemical charging, various positive or negative shifts of the G′ mode frequency have been reported. 6,15-17 Our detailed analysis of the dependence of the G′ feature on the electrode potential for several individual SWCNTs showed that the wide variety of the behaviors of the G′ feature is intrinsic to different (n,m) SWCNTs. Furthermore, our data on individual SWCNTs uncovered effects that were not previously observed in SWCNT bundles and are not explained by any current model. We also show here that the behavior of the G′ mode is very complex and further experiments must be performed to gain a more complete understanding of the observed phenomena. The results reported here contribute to a rationalization of the results obtained previously for SWCNT bundles,6,16-18 and serve to guide theorists in their development of models to explain these phenomena. Results and Discussion The scanning electron micrograph and an atomic force microscopy (AFM) image in Figure 1a and Figure 1b, respectively, show the morphology of our individual SWCNTs grown on a SiO2/Si substrate. The SWCNTs are parallel to each other, and they are up to several millimeters in length (see Figure 1a). The typical distance between two neighboring SWCNTs is larger than 50 µm. The spacing between tubes is important, since it determines the number of tubes measured simultaneously during acquisition of a Raman spectrum. Since the size of the focused laser spot was about 1 µm, our Raman measurement is almost always addressing just one individual SWCNT only. However, it is difficult to distinguish between one SWCNT and a small bundle containing two or three parallel SWCNTs, and thus the Behaviors of Raman G′ Mode of SWCTs J. Phys. Chem. C, Vol. 113, No. 5, 2009 1753 Figure 3. Dependence of the frequency of the G′ feature ωG′ on electrode potential for 11 different SWCNTs (identical symbols in different curves correspond to one tube, measured with two different Elaser values). The spectra are excited by 2.33 (a, i, k), 2.07 (b, f, h), 2.11 (c), 2.13 (d), 2.03 (e), 2.00 (g), 2.17 (j), 1.91 (l), or 1.76 (m) eV laser radiation. The error of the evaluation of the frequency is estimated as (2 cm-1. Figure 2. In situ Raman spectroelectrochemical data on an individual SWCNT in the range from -1.5 to 1.5 V (from bottom to top). The spectra are excited by 1.92 eV laser radiation. The bold spectrum corresponds to 0 V. The electrochemical potential change between adjacent curves is 0.1 V. The spectra are offset for clarity, but the intensity scale is the same for all spectra. The vertical solid line is a guide to the eye and corresponds to the position of the G′ phonon feature at 0 V. possible occurrence of such small bundles in our samples cannot be excluded. Some tubes in a small bundle are not detected in Raman spectra, because they are out of resonance, and thus mimics the behavior of an individual tube. Figure 2 shows in situ Raman spectroelectrochemical data measured on an individual SWCNT using 1.92 eV laser radiation. The electrochemical doping of a SWCNT shifts the frequency of the G′ mode. The original frequency of the G′ mode at 0 V (which is close to the open-circuit potential) is around 2620 cm-1. Increasing the potential to 1.5 V causes the frequency to downshift toward 2612 cm-1. On the other hand, decreasing the potential to -1.5 V results in a more pronounced downshift of the G′ mode to a frequency around 2600 cm-1. Previous experiments on SWCNT bundles also showed a change of the G′ mode frequency during electrochemical doping, but never showed a downshift for both positive and negative potentials.6,15-17 For positive doping of nanotube bundles an upshift of the G′ mode frequency has been found, while almost no shift was observed for negative doping using the 1.96 eV laser excitation energy.6 This behavior for bundled tubes is obviously different from our observation at the individual nanotube level (Figure 2). The interpretation of effects in nanotube bundles is complicated by the fact that many interacting tubes contribute to the actual acquired spectra and the doping level is inhomogeneous across the bundle. Nevertheless, the dominating argument for the potential-driven change of the G′ mode frequency was the hardening/softening of the C-C bond.6 On the other hand, it was found for SWCNT bundles that the frequency shifts are small with 488 nm laser excitation, but a strong upshift of the G′ mode with increasing electrode potential is found for the 568 nm laser excitation with small changes in electrode potential.17 This effect cannot be explained by hardening/softening of the C-C bond. Thus it was suggested that it is a consequence of the change of the shape of the highenergy TO phonon branch in the phonon dispersion of graphene. It was assumed, that the increased electrode potential leads to a flattening of this branch.17 This causes a change in the slope of the TO branch for lower laser energy excitations, while the “upper” part of this branch (accessed by higher energy excitation) remains unaffected.17 In other words, at lower energy excitation, the TO mode energy changes with applied potential, and therefore the frequency of the Raman G′ mode is changed.17 However, this model explains only the increase in the frequency of the G′ mode with increasing electrode potential. This is obviously in contrast to our experimental results at the single nanotube level, where ωG′ decreases with both increasing V and decreasing V (Figure 2). Note, in contrast, that for SWCNT bundles an upshift of the G′ mode frequency with increasing electrode potential is usually observed. Only for doping at negative potentials was a downshift sometimes found in prior work on bundle SWCNT samples.18 Considering only the results obtained for a bundled sample, it is impossible to identify the effects which are responsible for the G′ mode frequency variations with applied electrochemical potential shown in Figure 2. Various phenomena might be considered in interpreting the G′ mode frequency/potential dependencies: (i) the dependence on the particular nanotube (n,m), (ii) the dependence on the laser excitation energy and, consequently, on the particular Eii transition involved in the resonance enhancement, (iii) the changes of the C-C bond length as a result of doping, (iv) the dependence on the electrolyte solution and/or on other experimental conditions, and (v) the dependence on the difference between the laser excitation energy and the particular Eii of the nanotube that is in resonance with Elaser. To clarify these issues, we first accessed the behavior of the Raman spectra of different tubes using different laser excitation energies. Figure 3 demonstrates the large variety of behaviors 1754 J. Phys. Chem. C, Vol. 113, No. 5, 2009 Kalbac et al. found for the dependence of the G′ band frequency on electrode potential for different individual SWCNTs. For example tubes b, e, and f show a very weak or no dependence on electrode potential, much less than is typical for a bundle sample. Tubes d, i, j, and l exhibit a decrease of the G′ mode frequency with applying both negative and positive potentials as for the case of the tube in Figure 2. However, for tube d the effects were small, while for tubes i and l the effects were large. For tubes d and j the effects were symmetric with positive and negative potentials, while for tubes i and l the behaviors were strongly asymmetric for positive and negative potentials. On the other hand, tubes a and k show an increase of the G′ mode frequency with applying both negative and positive potentials. Tube m does not reflect the changes with a negative applied potential; rather the G′ mode frequency increases with increasing positive potential. Finally, tube h shows a strong decrease of the G′ mode frequency during negative doping. For positive doping of tube h, the behavior of the G′ mode frequency is more complex since it decreases up to a potential of about 0.8 V and then it again increases in going to a potential of 1.5 V. This richness in behavior requires detailed studies involving many (n,m) tubes to tease out the detailed dependence of this variety of behaviors on tube type, never seen before for bundled samples. Obviously ωG′ at 0 V is different for each curve. This reflects the dependence of ωG′ on the laser energy used for the excitation of the Raman spectra, on the diameter (and perhaps also the chirality) of the studied SWCNTs, and perhaps on whether the tube is metallic or semiconducting. The dependence of ωG′ on Elaser has been previously fitted to the equation ωG’ ) k1 + k2Elaser (1) where k1 ) 2420 cm-1 and k2 ) 106 cm-1/eV for individual SWCNTs on a SiO2/Si substrate.19 A similar value of k2 ) 96 cm-1/eV has been found for SWCNTs dispersed in SDS20 or for the outer tubes in DWCNT bundles (k2 ) 103 cm-1/eV).15 On the other hand, the narrower inner tubes of DWCNTs exhibited a dependence with a smaller slope.15 Here the values of k1 and k2 were 2448 cm-1 and 76 cm-1/eV, respectively, for the inner tubes in DWCNTs. The different slope for the inner tubes (relative to the outer tubes) is a consequence of the different values of Eii, which are involved in the resonant process for the inner and outer tubes.15,21 The dependence of ωG′ on Elaser is also demonstrated in Figure 3 for individual tubes. The curves measured on the same tube but at a different Elaser are shifted by a constant value, confirming the dependence of ωG′ on Elaser according to eq 1. Curve b in Figure 3 is upshifted by 5 cm-1 from curve c, and curve f is upshifted by ca. 8 cm-1 from curve g in Figure 3 (considering the values of ωG′ to be taken at 0 V, see Figure 2). Hence, the δ(ωG′)/δ(Elaser) values for curves f and g resemble a linear dependence of eq 1. On the other hand, the dependence δ(ωG′)/ δ(Elaser) for curves b and c has an opposite sign; that is, k2 is negative. This behavior may be indicative of a steplike character of the δ(ωG′)/δ(Elaser) dependence at the individual tube level (different Eii values are employed for Elaser ) 2.07 and 2.11 eV).19,22 Note that the difference in ωG′ for curves b and c reaches the resolution limit of the spectrometer. In order to evaluate the dependence of δ(ωG′)/δ(V) on excitation laser energy, we fitted the δ(ωG′)/δ(V) for each tube shown in Figure 3 for both negative and positive doping. For simplicity we used a linear fit even though some tubes showed a much more complex behavior. The results (not shown) confirm that there is no obvious dependence of δ(ωG′)/δ(V) on excitation laser energy for the individual SWCNTs studied in this work. The results were obtained for 11 different nanotubes using eight laser lines. Nevertheless, more in-depth study on a much larger number of tubes would be necessary to establish our preliminary finding on a sound experimental basis. It was shown recently that some of the features of the TG band (A1LO of semiconducting and metallic tubes and A1TO of semiconducting tubes) exhibit a change of frequency with applied gate voltage (V).23 The slope of this change is dependent on the diameter of the SWCNT.23 Therefore, a similar dependence can be suggested for the slope of the change of the G′ mode frequency with potential (δ(ωG′)/δ(V)). Unfortunately, it is very difficult to obtain Raman spectra for SWCNTs where both the RBM and the G′ modes appear with a reasonable intensity at the same time for the same tube. This complicates the evaluation of the tube diameter based on the frequency of the RBM band. Nevertheless, it was previously reported that the frequency of the G′ band (ωG′) scales with the inverse tube diameter d according to the equation ωG′ ) C1 - C2 /dn (2) where C1 ) 2708.1 cm-1, n ) 1, and the values of C2 in the literature vary from 35.4 to 67 cm-1 · nmn.15,24 (Note that the samples in these references are very different from each other. Our results lend credibility to the relatively broad range of values found in the literature for C2. The values of C1 and C2 are expected to be dependent on which Eii a particular tube is in resonance with, and whether the particular tube is semiconducting or metallic.) In addition, an alternative dependence with n ) 2 and C1 ) 2645 cm-1 was recently suggested by Cardenas et al.25 However, for our sample eq 2 gives the value of ωG′ up to 2675 cm-1 (Figure 3). Thus the original fit with C1 ) 2708.1 cm-1, n ) 1, and 35.4 < C2 < 67 cm-1 · nm15,24 seems to explain the experimental results more accurately. In our study here, we measured the tubes on a SiO2/Si substrate, which is similar to ref 24. However, the data in the latter reference were obtained using 2.41 eV laser excitation energy. Since ωG′ is dependent on Elaser, it is necessary to combine eqs 1 and 2 to estimate the diameters of the tubes. (For this purpose we used k2 ) 100 cm-1/ eV, C1 ) 2708.1 cm-1, C2 ) 35.4 cm-1 · nm, and n ) 1. The parameter k1 was calculated for each tube.) It is important to note that here we are neglecting the steplike behavior of δ(ωG′)/δ(Elaser)19,22 and we assume that C1 and C2 are the same for all tubes in our experiment. Thus the resulting tube diameters may exhibit some error for particular tubes. Nevertheless, we believe that this approximate procedure is sufficient to evaluate the dependence of the slope δ(ωG′)/δ(V) on tube diameter. In this way, we evaluated a plot of the slope δ(ωG′)/δ(V) for each of the 11 tubes for both positive and negative doping. Our experimental results showed no obvious dependence of δ(ωG′)/ δ(V) on tube diameter, which is a new result. Figure 4 shows the Raman spectra of the TG mode region for the same tubes that are presented in Figure 3 (at a potential 0 V). The broadening of the G- component (centered at ∼1550 cm-1) of the TG mode for curves e, f, g, and m indicates that these tubes are metallic. On the other hand, the spectra for the rest of the tubes do not exhibit such a broadening and thus they probably correspond to semiconducting tubes. The development of the TG band during electrochemical charging (not shown) confirms that curves e, f, g, and m in Figure 4 correspond to metallic tubes.26 (The intensity of the G- component is increased Behaviors of Raman G′ Mode of SWCTs J. Phys. Chem. C, Vol. 113, No. 5, 2009 1755 Figure 5. Dependence of the relative intensity of the G′ feature on the electrode potential for different SWCNTs (identical symbols for different curves correspond to one tube, measured with two different laser excitation energies). For all curves, the intensity at 0 V is set to unity. The spectra are excited by 1.92 (g), 2.00 (d), or 2.07 (a, c) eV laser radiation. Figure 4. Raman spectra in the TG mode region of tubes from Figure 3 (at a potential 0 V). The spectra are excited by 2.33 (a, i, k), 2.07 (b, f, h), 2.11 (c), 2,13 (d), 2.03 (e), 2.00 (g), 2.17 (j), 1.91 (l), or 1.76 (m) eV laser radiation. The arrows point to components of the TG band with the BWF line shape. as the potential is changed from 0 V to either 1.5 or -1.5 V.) The Raman bands below 1520 cm-1 and above 1650 cm-1 correspond to the electrolyte. The features of the electrolyte do not change with a change of the electrode potential, and they can therefore be easily recognized. It has been shown that the G′ mode splits for a particular nanotube due to a resonance enhancement employing both the incident and the scattered phonons.27 Closer inspection of the development of the decrease of the frequency of the G′ mode with changing electrode potential in Figure 3 shows that there is a small plateau, followed by a steeper decrease of the frequency, and finally there is a second plateau. Considering this, the development of ωG′ with changing potential could be alternatively rationalized in the following way. First, we consider that the spectral intensity of a particular tube is enhanced by the incident light via the resonance with the E44 transition for semiconducting tubes. Second, we assume that the change of the electrode potential can change the energy of Eii.6 Consequently, the change of the energy of E44 decreases the resonance enhancement. However, the electrochemical doping also causes changes in E33. Therefore for some tubes the next lower energy singularity E33 may come into resonance with the scattered light and a downshifted G′ feature appears in the spectra.27 This mechanism will be analogous to that reported previously for a steplike character of the δ(ωG′)/δ(Elaser) dependence on Elaser19,22 and is an effect that can only be studied at the individual nanotube level. The resonance with the scattered light always seems to be weaker than that with the incident light.27 In nanotube bundles, many tubes with different diameters are present and thus the G′ feature coming from the resonance with the scattered light can be easily hidden by stronger bands coming from the G′ feature excited by the incident light. The intensity of the G′ mode is expected to be dependent on the electrode potential. This change is attributed to the filling/ depletion of the states within the van Hove singularities. The electrochemical charging leads to a change of the Fermi level energy, and subsequently the electronic states are filled or depleted with electrons. The depletion of electronic charge is believed to suppress the resonance enhancement, and the spectrum is therefore bleached. In other words, the intensity should be constant if the Fermi level is below the energy of the van Hove singularity which is involved in resonance enhancement and should drop suddenly to zero when the Fermi level achieves the energy of this particular van Hove singularity. In contrast to these expectations, at the single SWCNT level we see cases where the intensity of the G′ mode bleaches continuously (as shown in Figures 2 and 5). A continuous bleaching of the spectral intensity was observed previously in nanotube bundles.6 In the latter case, the effect could be explained by the presence of many different tubes, with slightly different Eii values causing resonance enhancement. Furthermore, one can also argue that the interaction of the electrolyte counterion is different for a tube within the bulk of a bundle and with a tube at the surface of a bundle. These arguments are ruled out for individual nanotubes (Figure 5). Thus, for individual tubes, the bleaching of the spectrum is expected to be sudden and only at the potential corresponding to half the energy of a particular Eii. In our case this should occur at an electrode potential above 1 V. (In an ideal case, the electrode potential and the Fermi level shifts would be numerically identical. However, the experimental data show that the potential of 1 V corresponds to the Fermi level shift by 0.4-0.9 eV.28,29) In contrast to these expectations, the experimental results show a gradual decrease of the G′ mode intensity with increasing magnitude of the electrode potential (Figure 5). Therefore studies at the single SWCNT level show that a different explanation must be found. It was previously suggested for nanotube bundles that Eii changes with applied potential.6 In this case one would expect that, due to the change of the electrode potential, some of the tubes move out from the resonance window while other tubes move into resonance. Consequently, the overall spectral intensity should not change. However, for nanotube bundles the superposition of other effects mentioned above cannot be excluded, and thus the resulting dependence of the spectral intensity on electrode potential is a very complex issue. Here, in this study, the measured individual nanotubes have to match the resonance condition (otherwise there is no signal in the Raman spectra.) However, some of the tubes might not be in perfect resonance, and in principle, the shift of the van Hove singularity might 1756 J. Phys. Chem. C, Vol. 113, No. 5, 2009 lead to an improvement of the resonance condition. This effect could explain some of the “fluctuations” of the G′ mode intensity close to E ) 0 V in Figure 5. It has been shown previously that the substitutional doping of SWCNTs with boron or nitrogen causes a perturbation to the electronic structure.30 This is represented by new electronic states shifted from their unperturbed van Hove singularities.30 For example, it has been shown that the isolated nitrogen impurity forms a flat energy level lying inside the band gap.31 Implanting boron or nitrogen into nanotubes can furthermore lead to a change of the energy gap. We assume that electrochemical doping may cause analogous (but not the same) effects. The application of an electrode potential to SWCNTs leads to a stronger interaction of the SWCNTs with electrolyte ions which can result in a perturbation of the electronic structure of SWCNTs. Consequently, the density of states in van Hove singularities can be reduced. The “broadening” of van Hove singularities will extend the region where the SWCNTs can be in resonance, but the resonance effect would be weaker. This will explain a simultaneous resonance through E33 or E44 transitions and also a continuous bleaching of the Raman spectra during electrochemical doping. The interaction of the SWCNT with the SiO2 substrate may also contribute to the variety of behaviors of the G′ mode. A charge transfer between the substrate and the SWCNT together with an enhanced adsorption of air oxygen on the SWCNT due to the SWCNT-SiO2 substrate interaction are considered to be the important effects of the SiO2 substrate on the SWCNT.32,33 However, this charge transfer can be compensated by electrochemical doping, as has been shown previously.4,32 Our measurements are referred to the potential of the Ag pseudoreference electrode, and thus the doping level of an individual SWCNT is fully controlled in our experiment. Furthermore, the magnitude of the charge transfer between the SiO2 and SWCNT is very small compared to the potentials applied in our study. Other effects of the substrate such as substrate-induced strain or deformation of the nanotube shape are not expected to have a significant effect on our conclusions, since the effects observed in our study are fairly robust. Furthermore, we always follow a relative change of the G′ mode (for example, as a function of V or normalized to the G+ band intensity) and therefore the substrate should not influence the conclusions drawn from our experimental results. Therefore, we believe that the large variety of behaviors observed for the G′ mode is an intrinsic property of SWCNTs with different chiral indices (n,m). Only small perturbations of the G′ mode frequency and intensity are expected to be caused by interactions with the substrate and are also dependent on whether the SWCNT is semiconducting or metallic. Conclusion Raman spectroscopy and in situ Raman spectroelectrochemistry of the G′ mode were measured on individual SWCNTs, in order to evaluate the effect of electrochemical doping on this double-resonance feature. We show here that the G′ band electrochemical behavior of individual SWCNTs is sensitive to the nanotube structure (n,m) and that the details of this behavior are very complex. Some (n,m) SWCNTs exhibit a very weak dependence of the G′ mode on electrode potential, while other (n,m) SWCNTs show either a strong decrease or increase of the G′ mode frequency with either an increase or decrease in the potential. Such behaviors can only be studied at the single nanotube level. In the present work, we did not find any correlation between δ(ωG′)/δ(V) and Kalbac et al. either Elaser or nanotube diameter. On the other hand, we have observed a wide variety of behaviors, which require detailed study for many tubes in order to establish the differences between metallic and semiconducting tubes, and the varieties of each subset of metallic and semiconducting tubes exhibiting the various behaviors described in this paper. Our findings can rationalize contradictory results previously observed for the behavior of the frequency shift with changing electrode potential of the G′ mode in nanotube bundles. The nanotube bundles contain many tubes with different (n,m) indices. Using different excitation energies leads to a resonance enhancement of only a fraction of these tubes. The latter observation was previously interpreted as the dependence of the G′ band frequency shift with changing electrode potential on laser excitation energy. However, our results here show that the effect is instead caused by a large variety of behaviors of the G′ band frequency for individual (n,m) tubes presumably associated with their metallicity and tube type. In agreement with previous studies on SWCNT bundles, we have found that the intensity of the G′ mode is decreased in going from an electrode potential of 0 V to either +1.5 or -1.5 V. The relative change of the G′ mode intensity is monotonic and is qualitatively similar for all individual nanotubes that were studied. The monotonic change of the intensity is in contrast to our general expectation for observations at the individual tube level, and it was explained here by a broadening of the van Hove singularities as the electrode potential is changed in magnitude and sign. Last but not least, our data demonstrate the importance of the measurements at the single nanotube level, since the bundling effect together with the averaging of data measured for different nanotubes may lead to an incorrect interpretation of the observed effects. The variety of new phenomena that have been identified in the present work provides guidance for the development of theoretical models to account for the rich G′ band spectra that are observed in Raman spectroelectrochemical measurements, depending on the particular SWCNT (n,m) structure, on whether the tube is semiconducting or metallic, and on the magnitude and sign of the applied potential. Experimental Section SWCNTs were synthesized directly on a SiO2/Si substrate using Fe as a catalyst.14 The as-grown SWCNTs were contacted by Cr (5 nm)/Au (50 nm) evaporated on a part of the substrate, and the SWCNTs served as the working electrodes. The electrode potential was varied from +1.5 to -1.5 V. The cell was completed with a Pt counter electrode and an Ag wire pseudoreference electrode. The electrolyte medium was 0.1 M LiClO4 in propylene carbonate/poly(methyl methacrylate) (w/w 3:1) (Aldrich). Electrochemical doping was carried out potentiostatically (EG&G PAR potentiostat). The Raman spectra were excited by a dye laser (R6G dye, Coherent), a Ti:sapphire laser (Coherent), or a Kr+ laser (Coherent). The spectrometer was interfaced to a microscope (Carl-Zeiss, objective 100×). The diameter of the focused laser spot was about 1 µm. The AFM measurements were made with a Veeco Nanoscope instrument. Acknowledgment. This work was supported by the Academy of Sciences of the Czech Republic (Contract Nos. 203/07/J067, IAA400400804, and KAN200100801) and by the Czech Ministry of Education, Youth and Sports (Contract No. LC510 and ME09060). M.S.D. acknowledges support from NSF/ DMR-07-04197, H.F. and J.K. acknowledge the Materials, Structure and Devices (MSD) center, one of the five programs Behaviors of Raman G′ Mode of SWCTs in the focus center research program (FCRP). Work was carried out using the Raman facility in the Spectroscopy Laboratory supported by Grant NSF-CHE 0111370 and Grant NIHRR02594. Authors are grateful to M. Hofmann from MIT, Department of Electrical Engineering and Computer Science, for making the AFM measurements. References and Notes (1) Kurti, J.; Zolyomi, V.; Gruneis, A.; Kuzmany, H. Phys. ReV. B 2002, 65, 165433. (2) Thomsen, C.; Reich, S. Phys. ReV. Lett. 2000, 85, 5214. (3) Saito, R.; Jorio, A.; Souza, A. G.; Dresselhaus, G.; Dresselhaus, M. S.; Pimenta, M. A. Phys. ReV. Lett. 2002, 88, 027401. (4) Kavan, L.; Dunsch, L. ChemPhysChem 2007, 8, 975. (5) Kavan, L.; Kalbac, M.; Zukalova, M.; Dunsch, L. J. Phys. Chem. B 2005, 109, 19613. (6) Corio, P.; Jorio, A.; Demir, N.; Dresselhaus, M. S. Chem. Phys. Lett. 2004, 392, 396. (7) Stoll, M.; Rafailov, P. M.; Frenzel, W.; Thomsen, C. Chem. Phys. Lett. 2003, 375, 625. (8) Kalbac, M.; Kavan, L.; Zukalova, M.; Dunsch, L. AdV. Funct. Mater. 2005, 15, 418. (9) Kalbac, M.; Kavan, L.; Zukalova, M.; Dunsch, L. J. Phys. Chem. B 2004, 108, 6275. (10) Kalbac, M.; Kavan, L.; Dunsch, L.; Dresselhaus, M. S. Nano Lett. 2008, 8, 1257. (11) O’Connell, M. J.; Bachilo, S. M.; Huffman, C. B.; Moore, V. C.; Strano, M. S.; Haroz, E. H.; Rialon, K. L.; Boul, P. J.; Noon, W. H.; Kittrell, C.; Ma, J. P.; Hauge, R. H.; Weisman, R. B.; Smalley, R. E. Science 2002, 297, 593. (12) Kalbac, M.; Kavan, L.; Zukalova, M.; Pelouchova, H.; Janda, P.; Dunsch, L. ChemPhysChem 2005, 6, 426. (13) Jorio, A.; Saito, R.; Hafner, J. H.; Lieber, C. M.; Hunter, M.; McClure, T.; Dresselhaus, G.; Dresselhaus, M. S. Phys. ReV. Lett. 2001, 86, 1118. (14) Reina, A.; Hofmann, M.; Zhu, D.; Kong, J. J. Phys. Chem. C 2007, 111, 7292. J. Phys. Chem. C, Vol. 113, No. 5, 2009 1757 (15) Kalbac, M.; Kavan, L.; Zukalova, M.; Dunsch, L. Carbon 2004, 42, 2915. (16) Corio, P.; Santos, P. S.; Brar, V. W.; Samsonidze, G. G.; Chou, S. G.; Dresselhaus, M. S. Chem. Phys. Lett. 2003, 370, 675. (17) Rafailov, P. M.; Thomsen, C. J. Optoelectron. AdV. Mat. 2005, 7, 461. (18) Kim, Y. A.; Kojima, M.; Muramatsu, H.; Umemoto, S.; Watanabe, T.; Yoshida, K.; Sato, K.; Ikeda, T.; Hayashi, T.; Endo, M.; Terrones, M.; Dresselhaus, M. S. Small 2006, 2, 667. (19) Pimenta, M. A.; Hanlon, E. B.; Marucci, A.; Corio, P.; Brown, S. D. M.; Empedocles, S. A.; Bawendi, M. G.; Dresselhaus, G.; Dresselhaus, M. S. Braz. J. Phys. 2000, 30, 423. (20) Cardenas, J. F.; Gromov, A. Chem. Phys. Lett. 2007, 442, 409. (21) Hulman, M.; Pfeiffer, R.; Kuzmany, H. New J. Phys. 2004, 6, 1. (22) Dresselhaus, M. S.; Dresselhaus, G.; Jorio, A.; Souza, A. G.; Saito, R. Carbon 2002, 40, 2043. (23) Tsang, J. C.; Freitag, M.; Perebeinos, V.; Liu, J.; Avouris, P. Nat. Nanotechnol. 2007, 2, 725. (24) Souza, A. G.; Jorio, A.; Samsonidze, G. G.; Dresselhaus, G.; Pimenta, M. A.; Dresselhaus, M. S.; Swan, A. K.; Unlu, M. S.; Goldberg, B. B.; Saito, R. Phys. ReV. B 2003, 67, 035427. (25) Cardenas, J. F. Chem. Phys. Lett. 2006, 430, 367. (26) Farhat, H.; Son, H.; Samsonidze, G. G.; Reich, S.; Dresselhaus, M. S.; Kong, J. Phys. ReV. Lett. 2007, 99, 145506. (27) Souza, A. G.; Jorio, A.; Swan, A. K.; Unlu, M. S.; Goldberg, B. B.; Saito, R.; Hafner, J. H.; Lieber, C. M.; Pimenta, M. A.; Dresselhaus, G.; Dresselhaus, M. S. Phys. ReV. B 2002, 65, 085417. (28) Cronin, S. B.; Barnett, R.; Tinkham, M.; Chou, S. G.; Rabin, O.; Dresselhaus, M. S.; Swan, A. K.; Unlu, M. S.; Goldberg, B. B. Appl. Phys. Lett. 2004, 84, 2052. (29) Rafailov, P. M.; Maultzsch, J.; Thomsen, C.; Kataura, H. Phys. ReV. B 2005, 72, 045411. (30) Moradian, R.; Azadi, S. Physica E: Low-Dimens. Syst. Nanostruct. 2006, 35, 157. (31) Nevidomskyy, A. H.; Csanyi, G.; Payne, M. C. Phys. ReV. Lett. 2003, 91, 105502. (32) Shim, M.; Gaur, A.; Nguyen, K. T.; Abdula, D.; Ozel, T. J. Phys. Chem. C 2008, 112, 13017. (33) Minot, E. D.; Yaish, Y.; Sazonova, V.; McEuen, P. L. Nature 2004, 428, 536. JP808797C
© Copyright 2026 Paperzz