Disentangling Vacancy Oxidation on Metallicity-Sorted Carbon Nanotubes

TL;DR

This study uses first-principles calculations and XPS spectroscopy to identify and quantify oxygen-containing functional groups on vacancies in metallic and semiconducting carbon nanotubes, revealing differences in defect structures.

Contribution

It demonstrates how density functional theory can interpret XPS spectra to distinguish functional groups on oxidized vacancies in carbon nanotubes, advancing defect characterization methods.

Findings

Oxygen functional groups on vacancies are mainly epoxy, carbonyl, and ketene.

Metallic SWCNTs have 60% more O-containing defect structures than semiconducting ones.

O$_2$ does not contribute to the XPS O 1s spectra in oxidized SWCNTs.

Abstract

Pristine single-walled carbon nanotubes (SWCNTs) are rather inert to O$_2$ and N$_2$, which for low doses chemisorb only on defect sites or vacancies of the SWCNTs at the ppm level. However, very low doping has a major effect on the electronic properties and conductivity of the SWCNTs. Already at low O$_2$ doses (80 L), the X-ray photoelectron spectroscopy (XPS) O 1s signal becomes saturated, indicating nearly all the SWCNT's vacancies have been oxidized. As a result, probing vacancy oxidation on SWCNTs via XPS yields spectra with rather low signal-to-noise ratios, even for metallicity-sorted SWCNTs. We show that, even under these conditions, the first principles density functional theory calculated Kohn-Sham O 1s binding energies may be used to assign the XPS O 1s spectra for oxidized vacancies on SWCNTs into its individual components. This allows one to determine the specific functional groups or bonding environments measured. We find the XPS O 1s signal is mostly due to three O-containing functional groups on SWCNT vacancies: epoxy (C$_2$$>$O), carbonyl (C$_2$$>$C$=$O), and ketene (C$=$C$=$O), as ordered by abundance. Upon oxidation of nearly all the SWCNT's vacancies, the central peak's intensity for the metallic SWCNT sample is 60\% greater than for the semiconducting SWCNT sample. This suggests a greater abundance of O-containing defect structures on the metallic SWCNT sample. For both metallic and semiconducting SWCNTs, we find O$_2$ does not contribute to the measured XPS O~1s spectra.