Length dependence of the Raman spectra of carbon nanotubes

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2009.Includes bibliographical references.DNA-wrapping technology, combined with size-exclusion chromatography, have made possible the sorting of carbon nanotubes according to length. In particular, length sorted nanotube samples, with finite lengths approaching the exciton size, are now available. Hence, experiments that explore the finite size effects, in addition to length dependence, are now possible. Although there are many experimental and theoretical studies of the diameter dependence of carbon nanotubes, only a few studies of length-dependent effects exist. Raman spectroscopy is a powerful tool for the characterization of carbon nanotubes. Although Raman spectroscopy nominally probes the phonon scattering processes in carbon nanotubes, the technique provides information on a wide variety of phenomena and properties, including diameter and chirality, defects, and electron-phonon coupling. This thesis presents the measurement and analysis of Raman spectra in length sorted, DNA-wrapped carbon nanotubes. Three different samples of short nanotubes having lengths of 100 nm, 70 nm, and 50 nm, where the 100 nm tubes behave quite similarly to long nanotubes, were studied using several different values of laser excitation energy. We study the length and excitation energy dependence of various features, each reflecting different physical processes, in the Raman spectra of the length sorted nanotubes. Specifically, the G-band (which includes the G+ and G- peaks), D-band, G', G*, and RBM Raman features were examined. We have found that the spectra fall into two categories: spectra with a narrow G& peak, and spectra with a broad and asymmetric Gpeak which is commonly observed in metallic nanotubes. Although a mixture of semiconducting and metallic nanotubes are resonant for each laser excitation energy, we tentatively assign the spectra with a narrow G- peak as being from semiconducting tubes, and spectra exhibiting a broad, asymmetric G- peak to metallic tubes.(cont.) When the spectra are divided in this way, we have found that the properties of the various Raman peaks differ significantly between the two groups. We have found that the Raman spectra, especially for metallic nanotubes, varies for different spots on the same sample. This results in noisier data, for which the trends are not clear. We describe the use of the linear correlation analysis method, which allows correlations to be identified despite noisy data. For several of the Raman features, we have found that metallic nanotubes appear to be more sensitive to length than semiconducting tubes. From our analysis of the G-band, electron-phonon coupling seems to decrease for shorter nanotubes. The correlations between the different G-band parameters (frequency, linewidth, normalized intensity, and asymmetry) were identified. The D-band, which arises from a breakdown in selection rules brought about by finite size effects, was found to have a normalized intensity (relative to the G+ peak) that is inversely proportional to the nanotube length. We observe that the G' feature is comprised of two peaks, which is a result of two energy levels being simultaneously excited. Although the G* peak was weak and noisy, a few length dependent effects above the noise level were observed. Finally, we also observed some length dependence in the RBM peak.by Aurea Tucay Zare.Ph.D