Since their discovery [1] carbon nanotubes are a front line research topic. These of needle-like configurations, observed during intensive research work on fullerene C60,were found differently fascinated and known to have single, double and multi walled [1] [2] structures. Properties of these systems have been studied theoretically and experimentally for over decades. The theoretical method involved various techniques including simulations by means of different algorithms [3] . The experimental method, however, required sample preparation and characterization phenomenon [1] [2] [4] [5] . In a sample preparation arc discharge, Chemical Vapor Deposition (CVD) and Laser Ablation [6] methods have been utilized at most. Characterizations and typical analysis have been done by Raman Spectroscopy, X-Ray Diffiractometer (XRD), Transmission Electron Microscopy (TEM), Scanning Electron Microscopy (SEM), Atomic Force Microscopy (AFM), Energy Dispersion X-Ray (EDX), Scan- ning Tunneling Microscopy (STM), etc. depending on the type of the structure and morphology to be studied [7] .

The quality and quantity of the nanotubes is understood as to depend on the type of the discharging method, annealing temperature and time, refluxing temperature and time, system geometry, the electric current and voltage applied, and type of acids used for the reflux [8] - [10] . In most cases samples are prepared from commercially available graphite rods, mounted on electrodes of the discharger and kept few millimeters of distances apart before driving one towards the other, in certain gas or liquid environment [1] [6] [9] .

Following the realistic tight-binding band calculations by Hamada et al., 1992 [11] , experimental and theo- retical works have been reporting that these tubes can be either metals [12] or semiconductors [13] [14] . Elec- tronic band structure calculations have also predicted that the (n,m) indices determine the metallic or semicon- ducting behavior of CNTs [11] [15] . These nanotubes are understood as efficient sources of electron field emit- ters [16] [17] and enabled fabrication of remarkable varieties of field-effect transistors [18] - [20] for potential applications.

Studies indicate that, whether the product is single walled or multi walled depends on the amount of catalyst used [21] . Accordingly, the lesser the catalyst leads to multi walled carbon nanotubes production. This might result in the variation of sizes. Diameter of single walled carbon nanotubes (SWNTs) could vary up to 1.4 nm and millimeters of length [22] , and that of multiwalled carbon nanotubes (MWNTs) up to hundreds of nanometers width [23] almost independent of the preparation temperature [24] . It has been reported that production of MWNTs by carbon arc discharge method does not require any catalyst. The remaining techniques, however, in- volved metal doping and resulted in producing lesser crystalline and many more defects [25] .

In this research, we used the arc-discharge method and prepared carbon nanotubes in a deionized water. Structural analyses of the samples were done by spectra, diffraction and image pattern recording.

In sample preparation we used carbon graphite rods, of nearly 23 cm in length and 7.64 mm diameter, commercially available. These rods were cut into pieces and mounted on the electrodes of the arc used as anode and ca- thode, kept few millimeters apart in a chamber of deionized water cover, as shown in Figure 1, before the process.

A d.c. of 50 - 200 A driven by 40 V created a high temperature discharge between the two electrodes where high sparking and little smoke was observed. After frequent discharging the chamber is removed and the product soot along with water is transferred into a bigger beaker and kept covered with aluminum foil for about 6 hours until crude is formed at the bottom and then decanted. Subsequently, the crude is made open air dry at 100˚C for about 12 hours and the dried soot is collected for technical analysis.

The crystal purity and defect concentration of the graphite powder was tested by the mechanism of Raman spectra recording at room temperature, using RENISHAW-Raman equipment operating with Argon laser of one excitation and wavelength 514 nm.

The I_D/I_G ratio indicated that the graphite used contains large amount of defects, as illustrated in Figure 2(a). The collected data plots in Figure 2(b) and Figure 2(c) show that there is D-band (disorder induced phonon mode [26] ) at 1348 cm⁻¹ and strong peak G (graphite)-band at 1583 cm⁻¹ Raman shift for the as prepared sample and the D-band at 1349 cm⁻¹ and the G-band at 1582 cm⁻¹ for the annealed sample, respectively. Further analysis indicates that there is a G-peak due to a high production of monolayer graphene [27] at 2710 cm⁻¹. The increase in I_D/I_G is, therefore, because of that annealing successfully altered the CNTs, perhaps, increasing the number of defects on their side walls [28] . The increase in D-band frequency might be due to the chemical charge transfer under different temperature treatment and/or amorphous carbon content most likely from the destruction of the CNTs [29] . The D-band (sp³) is attributed as associated with vibrations of carbon atoms with dangling bonds in the terminal plane of disordered carbon/impurities whereas the G-band is related to the vibration of sp² bonded carbon atoms in a two-dimensional hexagonal lattice [30] . Figure 2(c) shows that there are D’-band peaks at 1614 cm⁻¹ and 1629 cm⁻¹ indicating the presence of randomly distributed impurities or surface charges in the graphene, resulting in splitting of the G-band in to G and D’-peaks. Further scrutiny shows that there is a G’- band (the second strongest after the G mode and the second overtone of the defect-induced D mode) at about 2708 cm⁻¹ and attributed as useful in determining the number of graphene layers [31] .

The X-Ray Diffractometer (XRD) D8 advanced, from Bruker A × S, of scan type locked coupled, scan angle 20˚ - 80˚ range, scan step 0.02˚, scan speed 3˚/min, max. power 40 kV/40mA, Cu tube, T/T horizontal, for which scan time is about 20 minutes for each sample and wavelength 1.5406 A˚ is used to identify the composition of the samples.

Accordingly, the peaks at about 26˚, 44˚, and 76˚ shown in Figure 3 corresponds to indices of graphite C(002), C(101) and C(110) planes, respectively, indicating the presence of CNTs in the sample and the scattering of incident beam by (002) faces [32] which always exists, as far as X-ray diffraction pattern of CNT’s are close to that of graphite [33] , in agreement with experimental results obtained earlier and JCPDF #751621 indications [34] .

The analysis was repeated, using another diffractometer, model name X’Pert PRO, company name PANalyti- cal (formerly known as Philips) with scan type continues, scan angle 10˚ - 90˚, step size 10,167/degree, time for step 20 sec, by CuKα₁ laser radiation, scan speed 0.1˚/sec, λ = 1.540598 A˚ wavelength, generator power setting 40 kV, and current of 40 mA, for conformation.

The data plot of the result, shown in Figure 4(b), also suggests that the reoccurrence of the peaks would come from the defects found in the raw material as far as reappeared in the unprocessed graphite powder, at about 28.2˚ for instance, and revealed by broadened line width at about 25.77˚ for graphite as well (see Figure 4(a)).

The EDX analysis indicates that there are Calcium (Ca) and Oxygen (O) constituents in the samples, perhaps, introduced during industrial preparation of graphite rod, as shown in Figure 5, and are known as introducing extra peaks at about 28˚, 47˚, and 56˚, and could not be removed by annealing. Cu peaks are due to grip on which the sample was deposited for analysis.

Scherrer’s formula [35] t = kλ/βcosθ_B which is derived from Bragg’s equation that has been utilized in determining crystallite inter planar spacing, is employed in estimating the nanotubes thickness t, where κ is the shape factor approximated to 0.9, β = Δ2θ × π/180˚ is the line broadening at half the maximum intensity (at full width half of maximum intensity, FWHM), λ is the X-ray wave length and θ_B is the Bragg’s angle.

One can understand, from Table 1, that the diffraction peaks at 2θ are slightly shifted as a result of further annealing (Figure 3(c)) perhaps due to formation of CNTs whose production has been overwhelmed by the defects. Moreover, the increase in thickness can be due to self-assembling of the CNTs.

Table 2 shows that sizes of the attributed defects could decrease with increase in the Bragg’s angle at 800˚C annealing. However, there are no well defined circumstances in the case of the pristine and the 400˚C annealed samples. It is worth mentioning that the size of those defects in the unprocessed graphite sample at about 28.2˚ is nearly 33.886 A˚. This indicates that size of the defects has increased due to thermal agitation and perhaps burning of some of the CNTs which are known to self categorize as also shown in Figure 8.

Images of the CNTs were collected using transmission electron microscopy (TEM) model 7500, 2 keV HITACHI, maximum magnification 6 × 10⁶ times and resolution 0.2 A˚, after sonication of the nanopowder suspending in ethanol and exposing to ultrasonic waves for 3 hr. As shown in Figure 6, the as prepared sample has contained puffy colored impurities stacked on the surface of the tubes and also scattered elsewhere. After open air annealing at 400˚C the density of these impurities and/or amorphous carbon [10] is reduced and the clarity of the CNTs is improved, as in Figure 7. This shows that further annealing to certain temperature limit may give better and more purified CNT products. The measured internal diameter of these tubes is known to vary nearly from 1.0 nm - 7.0 nm and the external diameter ranges between 6 nm - 26 nm, where their length extends to about 0.4 μm.

We have also further annealed the CNTs at 800˚C for 30 minutes and cooled back to room temperature. The color feature of the sample powder was found changed to grey and reduced in quantity, in agreement with pre-

viously obtained results [25] . According to the TEM images the CNTs are extremely clean and condensed in comparison to the lower temperature annealed samples shown in Figure 6 and Figure 7. There were large amounts of heavy sheets of graphene layers, amorphous carbon and/or other impurities and perhaps broken CNTs produced, as seen in Figures 8(b)-8(d). These implies that some of the CNTs might have burned and are unstable at higher temperatures [36] , suggesting reduction of the annealing temperature and/or time could resolve the scenario of obtaining high yield. Moreover, the amount of impurities scattered around the CNTs shown in Figure 6 are known to decrease with further annealing and the CNTs are progressively freed from those surface bonded agents. These CNTs are found in bundle rather than scattered as the graphene layers and/or amorphous carbon does as well. An interesting feature is, therefore, the separation of the system into different category. The CNTs bundled as in Figure 8(a); graphene sheets of edge thickness about 20 nm forced to be arranged in

order, as in Figure 8(b); micron sized amorphous system gathered and attached to one another as in Figure 8(c); and black spotted objects, thought as broken CNTs, of size 5 nm - 30 nm in different category as in Figure 8(d). These indicate that during further annealing structures of the same category get sorted and move to one side assisting enhancement of the purification process.

The prepared and purified CNTs in a deionized water environment are understood to have the same feature as those prepared in N₂, H₂, He or their mixture. The challenging scenario was finding the source of extra peaks observed during XRD analysis, which could have been resolved by EDX analysis that clearly shown that Ca, O and/or their compound existed in the sample. According to the TEM images, the CNTs are identified as to self- assemble further being thermally agitated with increase in temperature. The increase in I_D/I_G ratio can be attributed as indicator for structural defects due to annealing the CNTs and presence of impurities. It is also understood that elemental analysis should be done for the contents of the sample source (graphene) prior to any course of action in order to easily manage experimental procedures and fix treatment technique for the removal of defects. Finally, the water environment preparation of CNTs is understood as the most cost effective mechanism relative to those has been in use. Thermal agitation is also known to self-categorizing constituents in the sample; signifying complete removal of defects requires additional treatment technique.

Acknowledgements

We acknowledge the financial support from the C. V. Raman fellow ship for African researchers and Mr. Charanjit Singh, department of chemistry Panjab University, for his invaluable assistance.