Low Kinetic Energy Oxygen Ion Irradiation of Vertically Aligned Carbon Nanotubes

: Vertically aligned multiwalled carbon nanotubes (v-CNTs) were functionalized with oxygen groups using low kinetic energy oxygen ion irradiation. X-ray photoelectron spectroscopy (XPS) analysis indicates that oxygen ion irradiation produces three di ﬀ erent types of oxygen functional groups at the CNTs surface: epoxide, carbonyl and carboxyl groups. The relative concentration of these groups depends on the parameters used for oxygen ion irradiation. Scanning electron microscopy (SEM) shows that the macroscopic structure and alignment of v-CNTS are not a ﬀ ected by the ion irradiation and transmission electron microscopy (TEM) proves tip functionalization of v-CNTs. We observed that in comparison to oxygen plasma treatment, oxygen ion irradiation shows higher functionalization e ﬃ ciency and versatility. Ion irradiation leads to higher amount of oxygen grafting at the v-CNTs surface, besides di ﬀ erent functional groups and their relative concentration can be tuned varying the irradiation parameters.


Introduction
Carbon nanotubes (CNTs) are nanostructured material with high technological importance owing to their unique chemical and electronic properties. They have been reported as good candidates to replace silicon in different electronic devices due to their higher carrier velocity, nanoscale structure and relative low-cost large scale production [1]. Nevertheless, their inert surface limits their use in several applications, such as biological and gas sensor devices [2][3][4]. To tackle this issue, the CNTs surface properties can be altered through changes of their surface chemistry by functionalization. It has been shown that CNTs functionalization with oxygen groups can enhance their surface chemical reactivity [5,6] and change their hydrophobicity improving their dispersion in aqueous media [7,8], property that is of vital importance for their integration in biological systems. Moreover, oxygen groups can act as active sites for further functionalization increasing their potential applications in a vast variety of fields, such as drug delivery, bio imaging, water purification and catalysis, among others [9]. deposited on Si wafers and then, after its oxidation, a 1.5 nm Fe layer was deposited and heated to 300 • C for three hours to form nanoparticles, which are responsible for the CNTs growth. The Al and Fe layers were prepared by magnetron sputtering using a Quorum device. For the v-CNTs growth, the reactor was heated to 750 • C in atmospheric pressure under He flow (500 sccm). The catalyst was placed inside the reactor and H 2 flow (200 sccm) was introduced into the reactor for 20 min. After that, C 2 H 4 flow (60 sccm) was introduced into the reactor for 24 min. After the growth, H 2 and C 2 H 4 flows were switched off and a flow of He was used to clean the reactor (500 sccm). The outer diameter distribution of the v-CNTs obtained was 10-40 nm.
The low kinetic energy oxygen ion irradiation in v-CNTs was performed in a vacuum chamber system (PREVAC-541) using as an ion source a Tectra plasma source TPIS, where microwave energy is used to create gas plasma at the plasma cup from which ions are extracted. Microwaves with a frequency of 2.45 GHz were generated by a magnetron coupled to a coaxial feedthrough structure, which guided the microwaves into the plasma cup containing low pressure O 2 gas. The electrons generated in the plasma undergo cyclotron resonance motion increasing their perpendicular kinetic energy due to a magnetic quadrupole arranged around the discharge chamber. Subsequently, when the energized free electrons collide with the gas in the volume, they cause ionization of the gas if their kinetic energy is larger than the ionization energy of the atoms or molecules. The ion extraction optics was composed of two grids located at the open end of the plasma cup, a positive voltage was applied to the first grid to accelerate ions extracting them from the plasma out to the sample, the other grid had a negative voltage and assisted in the ion extraction and the beam current.
Oxygen ion irradiation was performed varying separately the irradiation time and the ion kinetic energy. For the samples functionalized with different irradiation time the ion kinetic energy was fixed at 1 keV and the time was varied from 1, 2.5, 5 and 10 min. The samples functionalized varying the ion kinetic energy had irradiation time fixed to 5 min, the different ion kinetic energies were 0.1, 0.25, 0.5, 1, 1.5 and 2 keV. For all samples the current induced by the ion irradiation energy was 10 µA, certifying that the number of ions reaching the samples were the same (Table 1). Table 1. Summary of irradiation parameters used for oxygen functionalization of vertically aligned multiwalled carbon nanotubes (v-CNTs), oxygen (at. %) and carbon (at. %) relative concentration obtained by X-ray photoelectron spectroscopy (XPS). The vertical alignment and the morphology of v-CNTs were characterized before and after oxygen ion irradiation by scanning electron microscopy (SEM) using a JEOL-JSM-7500F-Field Emission Scanning Electron Microscope operated at 15 kV and transmission electron microscopy (TEM) using a JEM-1011 Jeol Ltd. Tokyo, Japan.
The chemical composition of the v-CNTs was studied with X-ray photoelectron spectroscopy (XPS) using a VERSAPROBE PHI 5000 from Physical Electronics, Chanhassen, MN U.S. equipped with a monochromatic Al Kα X-ray source. The XPS analysis chamber is connected to the ion irradiation vacuum chamber system to avoid contaminations. C1s and O1s core level spectra were recorded for each sample with an energy resolution of 0.6 eV.
Raman spectroscopy was collected using a Micro-Raman system (Senterra Bruker Optik GmbH, Ettlingen, Germany) with a resolution of 3 cm −1 using as excitation source a laser with a wavelength of 532 nm and 20 mW of power.

Results and Discussion
The macroscopic morphology and alignment of v-CNTs were evaluated with SEM microscopy before and after the oxygen ion irradiation in order to estimate if disorder was induced by ions impacting onto the sample surface. We observed that neither the density nor the alignment of CNTs were affected by the ion irradiation treatment (Figure 1). The understanding of the dependency of the oxygen bonding configurations in v-CNTs on the different parameters used during ion irradiation is a key step for optimal controlling functionalization and, consequently tailoring CNTs properties for applications such as in sensing. X-ray photoelectron spectroscopy (XPS) is a surface technique that, besides providing information on the relative atomic concentration of the elements present on a sample, is also suitable to characterize changes in the bonding configuration of the atoms via the chemical shift, i.e., via the analysis of the shift in the binding energy of the core electrons of the different atoms composing a sample.
We first recorded XPS survey spectra to evaluate if the oxygen ion irradiation had grafted undesired elements other than oxygen at the CNT surface. Figure 2 shows typical survey spectra recorded on v-CNTs before and after 5 min of irradiation with oxygen ions with 1 keV of kinetic energy. We could observe in the spectrum recorded before the ion irradiation a high intensity peak centered at 284.4 eV, generated by photoelectrons emitted from the C1s electronic level, and a very low intensity peak at 532.6 eV associated to oxygen contamination during the v-CNTs synthesis, evaluated to be 1.7 at. %. After ion irradiation the relative intensity of the peak generated by C1s photoelectrons decreased and the intensity of the peak at 532.6 eV corresponding to O1s photoelectrons increased remarkably. No other elements were found in the samples.   Figure 3b): from 0.5 to 2 keV, the relative oxygen concentration remained almost constant (15.0 at. %) independent of the ion energy. As the number of oxygen ions arriving at the sample surface was fixed throughout the experiment, we could conclude that the ion kinetic energy in this range did not affect the rate of ions grafted at the CNT surface. The amount of oxygen ions implanted thus depended only on the treatment time. The second regime was observed for oxygen ion irradiation from 0 up to 0.5 keV (Figure 3b), it is clear that when CNTs were irradiated with ions with 0 keV kinetic energy, i.e., the grid supply at the extractor was set to 0 keV, the amount of oxygen grafted was lower. Under these conditions no specific ion kinetic energy was selected, and all types of species generated inside the plasma chamber were able to reach the sample surface, including neutrals species. It is important to mention that the distance between the exit of the plasma cup in the Tectra ion source and the surface of the sample was about 12 cm, thus it was a remote plasma treatment in which the species with short-life time recombine before reaching the sample surface [28]. Under these conditions instead of ion irradiation, functionalization occurred via a plasma treatment. The quantity of oxygen grafted during this remote oxygen plasma treatment was 8.0 at. % about half that produced by ion energy irradiation, with ion kinetic energies from 0.5 to 2 keV (Figure 3b). The increase in the grafting of oxygen for increasing ion kinetic energy could be associated to an increase in the probability of defect creation as discussed by Lehtinen et al. in 2010 [29]. These authors, using atomistic computer simulations based on analytical potential and density-functional theory models, showed that for carbon nanostructures the probability of defect creation during low-mass ion irradiation, had a sharp increase up to 0.5 keV kinetic energy followed by a gradual decrease. This shows that the functionalization efficiency using low kinetic energy ion irradiation was higher than in the remote plasma treatment, since the impacting oxygen species were more energetic increasing the probability of creating defect sites into the CNT lattice [30]. Additionally, the oxygen species shorter lifetimes increased the possibility for recombination into stable gas-phase species unlikely to react with the tubes. A detailed analysis of the C1s XPS spectrum gives information on the chemical configuration of the oxygen grafted in the graphitic network. Figure 4 shows a comparison between the C1s spectra recorded on the pristine v-CNTs and after 10 min of oxygen ion irradiation at 1 keV of ion kinetic energy and the result of the curve fitting reproducing each spectrum. The pristine C1s spectrum is asymmetric, due to a re-adjustment of the Fermi level caused by electron-hole interaction after the electron scattering. In order to reproduce this asymmetry, the fitting of C1s peak was made with the Doniach-Sunjic function [31]. The peak had a binding energy of 284.4 eV and was reported to be generated by photoelectrons emitted from sp 2 carbon atoms [11]. We also observed a low intensity broad peak centered at 290.9 eV, corresponding to photoelectrons that lose kinetic energy to π-plasmon excitations. To reproduce this peak a Gaussian-Lorentzian was used (Figure 4a).
The C1s XPS spectra recorded after v-CNTs irradiation with oxygen ions show a drastic change, widening in the range 285-290 eV, due to the bonding of oxygen to carbon atoms (Figure 4b,  supplementary). To reproduce the spectrum for the sample irradiated 10 min it was necessary to use six components, the two used in the fitting of pristine v-CNTs and four other symmetric peaks (Gaussian-Lorentzian). These components were centered at binding energies of 285.0 eV, 286.6 eV, 287.5 eV and 288.8 eV.
The component at 285.0 eV was assigned to sp 3 carbon bonding occurring at defects introduced in the carbon lattice by ion irradiation. This could include interlayer and intertube cross-linking defects, amorphization and carbon bonding to CO x species. Induced in a controllable way, these defects can cause beneficial changes in chemical and electronic properties of the CNTs, promoting their application in different fields [16]. The components at 286.6 eV, 287.5 eV and 288.8 eV were associated to carbon bonded to oxygen. The electronegative oxygen atoms induce a positive charge in the carbon atoms thus changing the electron screening of their nucleus. This increased consequently the binding energy of the electrons, explaining the shift to higher binding energies for the components in the C1s spectra. Using spin polarized density functional calculations within the local density approximation (LDA) for the interpretation of the C1s XPS peak of oxygen functionalized carbon nanotubes, Bittencourt et al. [12], assigned the component at 286.6 eV to epoxide groups (C-O-C), the component 287.5 eV to carbonyl groups (C=O) and 288.8 eV to carboxyl groups (-COOH).
In order to understand the effect of varying the ion irradiation parameters in the chemistry of the v-CNTs, we analyzed the C1s XPS spectrum of each sample ( Table 2, Table S1, Figures S1-S7). To corroborate our results, the analysis of O1s XPS spectra was also performed (Table S2, Figures S8-S17). The same number of C-O components were obtained in C1s and O1s analysis.   Figure 5 shows the variation in the relative area of each component used to reproduce the C1s peak as a function of the treatment time and the ion kinetic energy. The relative contribution of component C1 (graphitic carbon (sp 2 -C)) decreased from the 96.3% in pristine v-CNTs to 62.7% when the treatment time reached 10 min, indicating that the oxygen functionalization occurred through breakdown of π bonding in the sp 2 -C bond. The relative area of component C2 (sp 3 -C) increased to 20% after 10 min of treatment (Figure 5a). When the treatment time was fixed to 5 min and the ion kinetic energy was the variant parameter, we observed that the ion energy did not significantly affect the relative area of the graphitic carbon component (C1), remaining in a range from 69.7% to 72.0% (for the range of energies from 0.5 to 2 keV). Similarly, the area of the component C2 (sp 3 -C) remained in a range from 15.7% to 19.0%.
These results suggest that for the samples irradiated with ion kinetic energy varying from 0.5 to 2 keV, approximately the same number of ions interact with the carbon nanotubes. At 0 keV kinetic energy (a remote plasma treatment [32]), only 8.8% of the sp 2 carbon (C1) was transformed into sp 3 (C3) and 5.5% into oxygen-bound carbon species (C4), confirming that the remote plasma treatment generated less defects at the CNT surface than ion irradiation (Figure 5b), with much lower oxygen grafting at the CNT surface (Figure 3b).
It is interesting to compare the grafting of oxygen atoms for 0 keV (remote plasma functionalization) with the ion irradiation results (Figures 6 and 7b). In the case of remote plasma functionalization, the only C-O bonding observed comes from epoxide groups. These consist of single oxygen atoms bonded covalently to a sp 2 -C surface, forming covalent bonds with two carbon neighbors with the oxygen atom sitting above the C-C bond centre. This is the least disruptive oxygen functional group (indeed epoxide groups can be added to graphene simple trough ozone exposure with UV light [33]), since they do not require carbon dangling bonds, and hence associated hole or edge formation. This therefore confirmed that the remote plasma was only adding very low energy isolated oxygen atoms to the nanotube surface (probably primarily neutrals) with minimal damage and disruption of the carbon layers. Epoxide oxygen could also be removed through relatively gentle reducing processes.
In contrast, the ion-implantation results show high concentrations of carbonyl and carboxyl groups, demonstrating that the lattice was being disrupted and carbon dangling bonds created, which were being oxygenated. This was consistent with a picture where incoming charged oxygen ions were arriving with kinetic energies far above the knock-on threshold energies of the lattice carbon, resulting in carbon displacements and increased local damage. This was also visible in the production of sp 3 -C carbon atoms as collateral damage. Thus the two processes were complementary, allowing choice of oxygen chemical functional group addition depending on end-application requirements.  We could further understand the oxidation processes occurring during ion irradiation by exploring the change in relative concentration of different C-O species over time (Figure 7a). Just as for the plasma treatment, epoxide species start forming on the basal plane from the beginning of the treatment, and gradually increase in concentration over the time. In parallel there was an extremely initial rapid formation of carbonyl groups. This presumably involved functionalization of tube tips, which are much more chemically reactive than the tube sidewalls and represent preferential addition sites [34]. During this period there was a gradual increase in sp 3 -C indicating the ion irradiation process is beginning to damage the nanotube carbon network (Figure 5a). After 5 min the network was sufficiently damaged that holes and internal edges are formed. These were C=O and -COOH edge-terminated, giving rise to a corresponding increase in the C=O and -COOH C1s peaks in the XPS.
The carbonyl groups were intermediate states and could oxygen saturate, transforming to carboxyl groups, helping to explain the relative stability in carbonyl concentration and large increase in -COOH in the C1s peak from 5 to 10 min. This process would be kinetically dependent on the rate of damage site and hole formation (i.e., new dangling bond creation rate), and was also reflected in the ion energy study (Figure 7b). As the ion kinetic energy increased there was a steady decrease in epoxide concentration, primarily because at higher energies there is an increased chance of damaging the lattice and so inserting oxygen as other (carbonyl or carboxyl) forms. The slight decrease in the sp 3 -C component observed when the sample was irradiated with 0.1 keV ions could be associated to a preferential removal of physically adsorbed species (amorphous carbon and oxygen species including water) occurring during the synthesis and due to the exposure to ambient air.
Raman spectroscopy data shows variations in the ratio of relative intensity of the primary Gand D-peaks consistent with the interpretation we present here based on the XPS data, i.e., there was a general trend of increasing D to G-peak intensity ratio with both ion treatment time and energy consistent with an overall increase in disorder in the structure (see 343 Supplementary Figure S18 and associated supplementary discussion). Figure 8 shows the typical TEM images recorded on carbon nanotubes pristine ( Figure 8a) and carbon nanotubes irradiated for 5 min with oxygen ions with kinetic energy of 2 keV (Figure 8b) samples. The pristine nanotubes show a certain degree of disorder, typical of chemical vapor deposition (CVD) grown non-annealed multi-walled CNTs. The overall structure of CNTs was maintained after oxygen ion irradiation with no significant degradation such as amorphization and formation of amorphous carbon aggregates. However, high resolution TEM microscopy would be needed to expand the analysis of the creation of smaller defects such as vacancies, vacancy-aggregates and interlayer defects. While the pristine CNTs had closed tips, after oxygen ion irradiation the tips could be seen to be now open, evidence of oxygen tip attack and functionalization.

Conclusions
Different oxygen functional groups can be created at the surface of vertically aligned carbon nanotubes by low kinetic energy oxygen ion irradiation. We show that defect creation, the chemical nature and concentration of functional groups depended on the irradiation time and the ion kinetic energy, indicating that low kinetic energy ion irradiation was an optimal functionalization method that could be used to fine tune the type and concentration of functional groups grafted at the carbon nanotube surface. The versatility of the method was a very clear advantage in comparison to other functionalization techniques, because using low kinetic ion irradiation we could tailor the surface reactivity of carbon nanotubes specifically for different applications, in a clean way and employing very short treatment times.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3417/9/24/5342/s1, Figure S1. XPS analysis of v-CNTs after 1 min of ion implantation with 1 keV of ion kinetic energy, Figure S2. XPS analysis of v-CNTs after 2.5 min of ion implantation with 1 keV of ion kinetic energy, Figure S3. XPS analysis of v-CNTs after 5 min of ion implantation with 1 keV of ion kinetic energy, Figure S4. XPS analysis of v-CNTs after 5 min of ion implantation with 0.1 keV of ion kinetic energy, Figure S5. XPS analysis of v-CNTs after 5 min of ion implantation with 0.25 keV of ion kinetic energy, Figure S6. XPS analysis of v-CNTs after 5 min of ion implantation with 0.5 keV of ion kinetic energy, Figure S7. XPS analysis of v-CNTs after 5 min of ion implantation with 1.5 keV of ion kinetic energy, Figure S8. XPS analysis of O1s spectrum of v-CNTs after 1 min of ion implantation with 1 keV of ion kinetic energy, Figure S9. XPS analysis of O1s spectrum of v-CNTs after 2.5 min of ion implantation with 1 keV of ion kinetic energy, Figure S10. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 1 keV of ion kinetic energy, Figure S11. XPS analysis of O1s spectrum of v-CNTs after 10 min of ion implantation with 1 keV of ion kinetic energy, Figure S12. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 0 keV of ion kinetic energy, Figure S13. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 0.1 keV of ion kinetic energy, Figure S14. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 0.25 keV of ion kinetic energy, Figure S15. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 0.5 keV of ion kinetic energy, Figure S16. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 1.5 keV of ion kinetic energy, Figure S17. XPS analysis of O1s spectrum of v-CNTs after 5 min of ion implantation with 2 keV of ion kinetic energy, Figure S18. Raman spectrum of v-CNTs functionalized with low kinetic energy oxygen ion irradiation with different treatment time (a) and ion kinetic energy (b). Table S1. Relative area and standard deviation of the components present in the C1s XPS analysis spectrum, Table S2. Relative area and standard deviation of the components present in the O1s XPS analysis spectrum.