Purity and Defect Characterization of Single-Wall Carbon Nanotubes Using Raman Spectroscopy

We investigated the purity and defects of single-wall carbon nanotubes (SWCNTs) produced by various synthetic methods including chemical vapor deposition, arc discharge, and laser ablation. The SWCNT samples were characterized using scanning electron microscopy (SEM), thermogravimetric analysis (TGA), and Raman spectroscopy. Quantitative analysis of SEM images suggested that the G-band Raman intensity serves as an index for the purity. By contrast, the intensity ratio of G-band to D-band (G/D ratio) reflects both the purity and the defect density of SWCNTs. The combination of G-band intensity and G/D ratio is useful for a quick, nondestructive evaluation of the purity and defect density of a SWCNT sample.


Introduction
Evaluating the quality of single-wall carbon nanotubes (SWCNTs) is very important, both in basic research and industrial application.To evaluate quality, we must consider two independent parameters: purity and defect density.Purity can be defined as a content ratio of SWCNTs to impurities, and the defect density can be defined as the abundance of structural defects on the nanotube walls.Raman spectroscopy has often been applied for the purity evaluation because a Raman mode around 1350 cm −1 (the "D-band") is sensitive to structural defects in the graphitic sp 2 network typical of carbonaceous impurities, such as amorphous carbon particles [1].In previous studies, the intensity ratio of the tangential mode of SWCNTs (G-band) to the D-band was used to discuss the purity [2][3][4][5].Because a pure SWCNT is also thought to have considerable D-band intensity due to structural defects, however, the evaluation based on the G/D ratio has uncertainty as to whether the Dband intensity reflects the amount of impurity particles or the defect density on the sidewalls of SWCNTs.For example, when the D-band mainly reflects carbonaceous impurities in a sample, the G/D ratio becomes a good index of SWCNT purity.Meanwhile, when there are fewer carbon impurities in the sample, the G/D ratio can be used to discuss SWCNT defects.To resolve this uncertainty in the evaluation due to the double meaning of the D-band, a clearer scale reflecting either the purity or the defect density is required.
The Raman intensity of the G-band and radial breathing modes (RBMs) has been applied to the purity evaluation of SWCNTs [3,6,7].Recently, Itkis et al. reported that the G-band area is proportional to the relative purity, as determined by optical absorption spectroscopy, for samples produced by arc discharge [3].In this paper, we show that the G-band peak intensity around 1593 cm −1 serves as a good index for the purity of as-grown SWCNTs produced by various synthetic methods, such as chemical vapor deposition (CVD), arc discharge, and laser ablation, even though their mean diameters differed.SWCNT purity was characterized using scanning electron microscopy (SEM) and compared to the G-band intensity and the G/D ratio of the samples.Quantitative analysis of SEM images suggested that the G-band intensity obtained from 2.31 eV laser excitation is associated with the purity of SWCNTs with mean diameters ranging from 1.0 to 2.0 nm.An improved method of purity and defect evaluation using Raman spectroscopy is proposed.

Experimental
The SWCNT samples used in this work and the measurement conditions for the purity evaluation were as follows.The six SWCNT samples are referred to as (1) HiPco (raw material produced by the HiPco process, lot no.R0546, Carbon Nanotechnologies Inc.), ( 2) CoMoCAT (purified material produced by the CoMoCAT method, Southwest Nanotechnologies Inc.), ( 3) Meijo (raw material produced by arc discharge, AP-J grade, Meijo Nano Carbon Inc.), ( 4) Carbolex (raw material produced by arc discharge, Carbolex Inc.), ( 5) Laser (raw material produced by laser ablation in our laboratory), and ( 6) DIPS (raw material produced by direct injection pyrolytic synthesis (DIPS), Nikkiso Co., Ltd.).Raman spectra were measured in the back-scattering geometry using a single monochromator with a microscope (LabRam Aramis, Horiba Jobin Yvon) equipped with a charge-coupled device detector and a notch filter.The sample was excited by the continuous wave second harmonic of an Nd:YAG laser at 2.31 eV (532 nm).To avoid laser heating, a 0.1 mW laser beam was focused onto the sample using an objective lens (×10).Raman spectra were obtained by averaging 20∼30 spectra obtained from different locations on the sample.All Raman measurements were carried out under the same conditions for all samples to maintain high uniformity of the intensity.Thermogravimetric analysis (TGA) profiles were recorded from room temperature to 900 • C in air flow (50 ccm) at a heating rate of 10 • C/min with a microthermobalance (TGA-50, Shimadzu).Samples weighing 1∼2 mg were used for the TGA measurements.The corresponding SEM images of the samples were obtained using JEOL JSM-7500F, operated at 15 kV.

Results and Discussion
The purity of the samples was first characterized quantitatively using SEM, which provided visual information of the ratio of SWCNTs to carbonaceous impurities.The SEM images and corresponding image analysis results of the samples are shown in Figure 1.We presumed that Fiberlike areas correspond to bundled SWCNTs, while the other particles and lumps correspond to carbonaceous impurities.The area counting of SWCNTs and carbonaceous impurities in the SEM images was carried out using image-processing software, ImageJ (NIH; http://rsb.info.nih.gov/ij/), as shown in Figure 1.The ratio of SWCNT area to particle area was assumed to relate to the purity of SWCNTs.The results for each sample are presented in Table 1 and Figure 2(a).Image analysis determined that the DIPS sample possessed the highest purity, and the HiPco sample had a higher purity than the Meijo, Laser, and Carbolex samples.Image analysis of the as-purchased CoMoCAT sample was unsuccessful because the SWCNTs were fully covered with carbonaceous impurities, probably due to the purification process.As a result, the SEM image could not be separated into SWCNT and impurity regions for analysis.The Raman and TGA results from the CoMoCAT sample were nevertheless included for reference.
It is well known that the G-band of SWCNTs shows multipeaks around 1580 cm −1 .For purity evaluation using Raman spectroscopy, the G-band peak around 1593 cm −1 , which derives from the longitudinal optical (LO) phonons of semiconducting SWCNTs [8], was used for two reasons: First, the G-band intensity is less sensitive to excitation laser energy than the RBM intensity, because of a loose resonance condition due to the large phonon energy.Second, a recent theoretical study predicted that there is no significant diameter dependence of the G-band intensity for the LO phonon of semiconducting SWCNTs, while the RBM intensity is more sensitive to the diameter and chirality for SWCNTs with diameter of 0.8∼2.0nm [8].The overtone of the D-band, the so-called G -band, was also predicted to depend on the chirality [9].Thus, the G-band around 1593 cm −1 is more appropriate than the other Raman modes for the comparison of purity of SWCNTs with different diameter distributions.
Raman spectra of the samples are presented in Figure 3.In the spectra, radial breathing modes (RBMs), the D-band, and the G-band were observed between 100∼400 cm −1 , 1250∼1350 cm −1 , and 1500∼1600 cm −1 , respectively [1,10].The average nanotube diameter of each sample was estimated using the relation of RBM frequency ω RBM to nanotube diameter Dt, Dt (nm) = 227/ω RBM (cm −1 ) [11].The Gband and the D-band intensities were obtained from their maximum peak counts.The G-band intensity and G/D ratio are shown in Table 1 and Figures 2(c) and 2(d).
In Figures 2(b) and 2(c), it can be seen that there was poor correlation between the G-band intensity and the G/D ratio.For the DIPS sample, which had the highest purity of the samples used, according to the SEM image analysis, both the G-band intensity and the G/D ratio were the highest of all samples.However, the G/D ratio of the Laser sample was 2.4 times higher than that of the HiPco sample, while the G-band intensity of the Laser sample was only 0.7 times of that of the HiPco sample.In the previous study, a linear relationship between the G/D ratio and the G-band was observed for samples produced by arc discharge [3].The present results show that such a relationship does not hold true for samples produced by different synthetic methods.
In contrast to the poor relationship between the G/D ratio and G-band intensity, the G-band intensity was well associated with the SWCNT purity estimated by SEM image analysis, as shown in Figures 2(a) and 2(b).Usually, asproduced SWCNT samples contain carbonaceous impurities, such as graphitic and/or amorphous carbon particles.Because both the SWCNTs and carbonaceous impurities have the same π electron system, the carbonaceous impurities also exhibit a black color and absorb visible light similarly to the SWCNTs.However, their contribution to the G-band Raman intensity is significantly different.Actually, the Raman signal from SWCNTs in a raw soot is about 30 times higher than that of carbonaceous impurities due to a resonance effect [3].Thus, it is reasonable to suppose that carbonaceous impurities act as an optical absorber and any observed difference in the G-band intensity can be mainly attributed to a difference in the amount of carbonaceous impurities present.
The G-band intensity may also be affected by additional factors such as excitation laser wavelength and nanotube diameter.It is known that the Raman intensity of SWCNTs is significantly enhanced by a resonance effect when the excitation energy matches the absorption bands of SWCNTs [12].The optical transition energy is inversely proportional to the nanotube diameter [13].According to the relationship between the diameter and optical transition energy of SWCNTs [13], the measured Raman spectra were on the resonance of the E 33 or E 44 optical transition of semiconducting SWCNTs, except for the CoMoCAT sample.Because the E 33 and E 44 transitions have broader spectral features than the E 22 transition [12,14], under the specific laser excitation, the difference in resonance conditions for each chirality is not significant.Thus, many types of SWCNTs can be excited by a single-wavelength laser.The resonance of semiconducting SWCNTs was confirmed by the sharp G-band peak around 1593 cm −1 .Consequently, the difference in the resonance effect and nanotube diameter did standard sample that gives a stable, uniform purity.As shown in Figures 1(a) and 3(a), the DIPS sample had extremely high purity and a highly stable G-band intensity.In this study, we used the DIPS sample as the standard to obtain the relative purity of the other samples.Although we primarily discuss as-grown samples, it is noteworthy that the Raman intensity of SWCNTs is sensitive to their aggregation [15] and to charge transfer, such as that induced by molecular adsorption [16].Because of these effects, it is important to monitor the purity of SWCNTs at each purification step.In the purification process, we often use acid, which easily induces charge transfer [16], and after the purification, SWCNTs tend to form thick bundles or sometimes aggregate with impurities.To avoid misleading results, it is necessary to check for aggregation of SWCNTs by SEM and to remove adsorbed chemicals from the purified materials, using vacuum annealing or some other process.In the present study, therefore, we have not used "purified samples" but "as-grown samples" which have a scarce charge transfer and randomly oriented bundles with the size from few nanometers to several tens nanometers.We note that these samples do not have specific orientations of nanotube which should be avoided because the resonance Raman intensity of SWCNTs has strong polarization dependence to the nanotube axis.
Once the purity of a sample is assessed from the G-band intensity, the G/D ratio can be used to discuss the relative abundance of SWCNT defects.The mismatch between the purity (Figure 2(a)) and the G/D ratio (Figure 2(c)) was caused by the difference in the main contribution to Dband intensity for each sample.It is known that the Dband derives from the structural defects of the graphitic sp 2 network in both SWCNTs and carbonaceous impurities, so the simple use of only the G/D ratio is not suitable to determine purity and quality.The HiPco sample, for example, possessed a higher purity than the Laser sample but had a lower G/D ratio.This suggests that the HiPco sample had more defects than the Laser sample.The combination of the G-band intensity and the G/D ratio can be highly useful to determine both the purity and the defect density, which has been difficult to accomplish by means other than Raman spectroscopy.
Raman and SEM were used for purity evaluation and could be associated with the ratio of SWCNTs and carbonaceous impurities in a sample.On the other hand, TGA was used to measure the content ratio of noncarbonaceous impurities in a sample by monitoring the weight decrease of samples during combustion in air.TGA profiles of the samples are presented in Figure 4.The weight ratio of carbon in the samples is shown in Figure 2(d).It can be seen that the weight ratio was different from the purity evaluated by SEM.Because the density of metal catalyst is much higher than that of carbon materials, evaluation of the weight ratio of the metal impurities to carbonaceous material is difficult, even by SEM image analysis.These methods complement each other.Thus, we conclude that it is better to combine Raman, SEM, and TGA for SWCNT purity evaluation and to compare the results in accordance with their intended use.

Conclusion
A comparative study of purity and Raman intensity was carried out for SWCNTs produced by various synthetic methods.The detailed comparison of Raman intensity and SEM observations suggested that the G-band intensity is a better indicator of SWCNT purity than the G/D ratio.The G/D ratio is not suitable for determination of the purity of various samples produced by different synthetic methods.The present results indicate that a combination of the G-band intensity and the G/D ratio is useful to evaluate the purity and defect density of SWCNTs.The evaluation conditions employed in this study can be applied to SWCNTs with an average diameter of 1∼2 nm, which covers most commercially available SWCNTs.If a consistent standard sample were available, evaluation based on the Gband and the D-band intensities could provide an easy means of determining the quality of SWCNTs for all users and suppliers of SWCNTs.

Figure 2 :
Figure 2: (a) relative area of SWCNTs in an SEM image, (b) normalized G-band Raman intensity, (c) G/D ratio, and (d) weight ratio of carbon in a sample for DIPS, HiPco, Meijo, Laser, Carbolex, and CoMoCAT samples.

Table 1 :
Relative area of SWCNTs in a SEM image, R Area (%), weight ratio of carbon to sample, R Carbon/Sample (wt%), average diameter of SWCNTs, Dt (nm), normalized G-band Raman intensity, I G /I DIPS , and the G/D ratio, I G /I D , for DIPS, HiPco, Meijo, Laser, Carbolex, and CoMoCAT samples.Sample R Area (%) R Carbon/Sample (wt%) Dt (nm) I G /I DIPS I G /I D