Controlled Oxidation of Single-Wall Carbon Nanotubes: A Raman Study

Oxidation of single wall carbon nanotubes using H2O2 is a common purification and tube opening procedure. We studied the effect of oxidation in well controlled conditions on SWNT samples with different mean diameters, prepared by laser-ablation, CVD and the HiPco process. Detailed multifrequency Raman spectroscopy evidences that oxidation damage depends strongly on the mean tube diameter, damage occuring to small tubes first and successively for larger diameter tubes. We use the peapod filling of the above samples as a control to what extent the tubes are open to the C60 molecules. INTRODUCTION Related to the synthesis method, SWCNT samples contain inevitable catalytic particles as well as of non-desired carbon compounds. SWCNTs are known to be resistant to oxidation better than any other carbon modifications or metallic particles. This inspires the purification of SWCNTs with oxidating treatment such as refluxing in H2O2 or heat treatment in air. Naturally, a reasonable balance have to be achieved when non-wanted side products or catalytic particles are nominally removed and the desired SWCNT sample remains yet in sufficient abundance. Another important aspect of the oxidation is the opening up of the SWCNTs for the environment thus making the encapsulation with materials such as alkali halides or C60 possible 1 . Here, we report a systematic gravimetric and Raman study of oxidation of different SWCNT samples using H2O2. Raman spectroscopy and in particular multifrequency Raman measurements have been proven to be crucial in the characterization of several properties of SWCNTs 2 . The level of oxidation is controlled by the dilution of the H2O2 aqueous solution. We found a strong dependence of oxidation on the nominal tube diameter of the samples, the small tubes being oxidized first, most probably from the tube ends. EXPERIMENTAL We studied 3 different SWCNT samples from commercial sources: laser ablation, LA, (Tubes@Rice, Rice University, Houston, Texas), CVD (Nanocyl, Namur, Belgium) and HiPco (Carbon Nanotechnolgies, Houston, USA). The purities as provided by the manufacturers are summarized in Table 1. Oxidation was studied by 2 hours refluxing of the SWCNT in diluted H2O2 aqueous solutions. After refluxing, the Downloaded 28 Nov 2007 to 131.130.1.18. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/proceedings/cpcr.jsp material was filtered and the resulting bucky paper was dried at 140 o C in air. Comparability and reproducibility was assured by the use of the same amount of starting SWCNT material, 15 mg, and the same volume for the aqueous solution, 30 ml, for all treatments. The unit of treatment is defined as 1mg SWCNT in 1 ml of 30 % aqueous H2O2 and 1 ml of distilled water. Numbers above unity are achieved by repeating the refluxing, filtering and drying steps. Peapod filling was done following the step in Ref. 3. The peapod concentration in the resulting materials was determined from the Raman signal of the C60 Ag(2) following Ref. 4. TABLE 1. Properties of the studied SWCNT materials Material Purity (%) Mean diam. (nm) Peapod concentration (%) HiPco 70 0.98 0 CVD 70 1.20 10 LA 15 1.34 30 RESULTS AND DISCUSSION Figure 1. shows the weight loss of the SWCNT materials after the described oxidation procedure. Weight loss is generally considered as a poorly controllable quantity since SWCNT materials are known to absorbe an ill defined and different amount of solvents. This originates in the different morphology of the SWCNT bundles. In order to reduce systematic errors related to this, we performed a control experiment, where no H2O2 was added to the distilled water. As shown in Fig. 1., the final and starting masses are in a close agreement, which supports the validity of the current gravimetric studies. A surprising observation in Fig. 1. is the complete dissappearance of the HiPco material at treatment unit 1, whereas the LA and CVD compounds seem to tend to a constant value of the final to initial mass ratio. This may originate in two facts: i.) the overally smaller diameters of the HiPco material may be more susceptible for the oxidation treatment than the larger diameter tubes present in the LA and CVD material (see. Table 1.); ii.) it is also possible that a different morphology related to the catalytic particles and the SWCNT may give rise to the difference. This later proposal is based on the fact the for the LA and CVD grown 0 2 4 6 0.0 0.2 0.4 0.6 0.8 1.0 f in a l/ s ta r t m a s s ( m g /m g ) V(H2O2 aq.)/ mass(SWCNT) (ml/mg) FIGURE 1. Mass loss of SWCNT samples after H2O2 treatment ( : HiPco, : laser ablation, : CVD). Dashed lines are guides to the eye. Downloaded 28 Nov 2007 to 131.130.1.18. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/proceedings/cpcr.jsp materials, the SWCNTs are known to grow out of larger catalytic particles, whereas the HiPco based SWCNT are known to be attached to smaller catalytic particles as well as to encapsulate the catalyst itself. The close presence of catalyst particles may catalyze the decay of H2O2 into the nascent oxygen reaction agent thus helping the oxidation of nanotubes. In what follows we focus our attention on the HiPco material in order to clearify the full dissappearance of the material with treatment. The peapod concentration attests to what level the SWCNT samples are opened, and whether the tube diameters are large enough to accommodate C60. Our results for the studied samples are summarized in Table 1. Clearly, even when significant number of tube openings are present, the HiPco tubes are too small for a detectable level of C60 encapsulation. Figure 2. shows the RBM mode region of the Raman spectra of the HiPco sample with the oxidation steps defined on Figure 3a. at λ= 488 nm. Changes associated with the treatment are observed: RBM lines at higher Raman shifts vanish with increasing treatment. SWCNT with larger Raman shifted RBM lines correspond to the thinner tubes as νRBM 1/d, where d is the tube diameter. The evident disappearance of thin SWCNT in the HiPco sample is quantitatively described when the mean tube diameter, d, and its variance, σ, is calculated assuming a monomodal distribution of the tube diameters following Ref. 5. Figure 3b. summarizes the data at 488 and 647 nm: the mean tube diameter gradually decreases that is accompanied by the narrowing of the diameter distribution function. This is understood as oxidation happening to the smallest nanotubes first. A simple calculation has shown that the shifting of d to higher values and the narrowing of the distribution can account for the ~80 % weight loss observed for the ‘6’ treatment. The Raman D and G modes also hold significant information about the oxidation damage. In Figure 4a. and b. we show the Ramand D and G modes at λ=488 nm. The 100 200 300 400 500 R a m a n s ig n a l ( a .u .) Raman shift (cm -1 ) T r e a tm e n t