662
Origin of Highly Efficient Iodine Doping in Carbon Nanotubes Studied By Raman Spectroscopy and First-Principles Calculations

Monday, 29 May 2017: 15:20
Churchill B1 (Hilton New Orleans Riverside)
A. Zubair (Rice University), D. Tristant, C. Nie (Université Fédérale de Toulouse-Midi-Pyrénées), D. E. Tsentalovich, R. J. Headrick, M. Pasquali (Rice University), E. Flahaut, M. Monthioux (Université Fédérale de Toulouse-Midi-Pyrénées), J. Kono (Rice University), I. Gerber, and P. Puech (Université Fédérale de Toulouse-Midi-Pyrénées)
Individual carbon nanotubes (CNTs) are highly conducting (both electrically and thermally), mechanically strong, and ultralight. These characteristics make them promising for use in various applications, including renewable energy, electronics, and aerospace. However, in macroscopic form, i.e., in the form of CNT bundles, these outstanding properties mostly vanish. Recently, fibers consisting of iodine-doped aligned CNTs with conductivities close to that of copper have been produced. Many studies have assumed polyiodide ions, such as tri-iodide (I3-) and penta-iodide (I5-), to be the origin of doping, but the observed large conductivity values cannot be explained with only polyiodides. Here, we quantitatively determine the doping level due to iodine and estimate the number of activated conductive channels in the double-wall CNTs (DWCNTs) that form the fibers. The CNT fibers used in this study were produced by the wet-spinning process, during which the nanotubes were doped with chlorosulfonic acid. The fibers were annealed for dedoping and then were kept in an iodine vapor chamber for 24 hours for doping. To quantify the conductivity increase due to iodine doping, we measured the current carrying capacity of the fibers. To determine the doping level, Raman experiments were performed at room temperature with 568 nm excitation on both iodine-doped and dedoped fibers. The figure shows G-band Raman spectra for (a) an iodine-doped DWCNT fiber and (b) a dedoped DWCNT fiber. These Raman spectra were fit with four Lorentzian peaks using the decomposition method developed from multiwavelength Raman spectroscopy studies on doped samples and high-pressure Raman experiments. These four peaks represent the upper (G+) and lower (G-) branches of the inner (Gi) and outer (Go) tubes, characteristics of the G-band of DWCNTs. For the iodine-doped fiber, Go-, Gi+ and Go+ are upshifted compared to the dedoped fiber. The G-band shift associated with the inner tube is due to the lattice contraction of the outer tube. From these shifts, it is possible to estimate the charge transfer per carbon atom, fC, to be 0.029 and the average Fermi level shift to be 1 eV. Due to the high level of doping, the number of conductive channels increases. To estimate the number of conductive channels, rough estimation was possible using Ei (eV) = 0.38i/d(nm) (for metallic tubes i = 0, 3, 6, ..., and for semiconducting tubes i = 1, 2, 4, 5, 7, 8, ...). Before doping, a statistical analysis gives 1/3 metallic (2 channels) and 2/3 semiconducting (0 channel) for inner and outer tubes, which suggests that the average number of conductive channels for a DWCNT is 4/3. As the outer diameter is ~2 nm, by moving the average Fermi level by 1 eV, we obtain i = ~5, which indicates that the number of conductive channels for a doped DWCNT is 26/3 on average. This conductance improvement is significant. By first-principles calculations based on the density functional theory (DFT), using the Vienna ab initio simulation package (VASP), the structural and electronic properties of both CNT systems were obtained. The plane-wave basis set cutoff energy was set to be 400 eV with a Gaussian smearing method of 0.005 eV width, in order to assure well-converged total energy and force values. All the atoms were allowed to relax until the maximum of forces acting on them became smaller than 0.01 eV/Angstrom. After optimization, the charge transfer of isolated CNTs and CNT bundles, respectively, was determined using Henkelman's group’s program for Bader charge analysis. This calculation proves that two I2 in contact with CNTs can interact to form metastable I3- and I- with a large charge transfer to the carbon nanotube, considerably increasing the doping to a level compatible with experimental observations. Finally, another set of samples were prepared with iodine inside the DWCNTs in order to clearly observe the arrangements of the iodine atoms. We observe, through transmission electron microscopy (TEM), isolated I adjacent to I3 species. Thus, large charge transfer due to I- and detection of I- through TEM explains the high conductivity achieved through iodine doping.