Production of carbon nanotube yarn from swirled floating catalyst chemical vapour deposition: a preliminary study

The SFCCVD reactor employed in this study was designed and fabricated as described in Iyuke [18]. It comprised of a 70 mm by 820 mm vertical quartz plug flow reactor in a furnace. The reactor is equipped with PT-100 thermocouples to indicate the temperature and a temperature controller to regulate the temperature. In addition, the system is equipped with a set of four rotameters and a flow controller to measure and control the flow of gases into the reactor. A 5-coil turn swirled mixer was fixed to the bottom end of the reactor to enhance the mixing of ferrocene aerosol, carbon precursor and carrier gas before entering into the reaction zone. An electric motor with spinning rate of 1400 rpm was attached to the top of the reactor to spin the produced CNT into yarns (see figure 1 for schematic illustration of the reactor).

Zoom In Zoom Out Reset image size

Nitrogen gas (99.999%), argon gas (99.999%) and acetylene gas ($\geqslant 99\%$) were purchased from Afrox (South Africa). Ferrocene (98%), used as the catalyst, was obtained from Sigma Aldrich (South Africa). All the materials were used as supplied without any further purification.

Before the production of yarn using the SFCCVD, the system was purged with nitrogen gas to remove any contaminants and also to check for gas leaks. The SFCCVD reactor was then heated up to a desired temperature while argon was passed through the system to create inert environment for the reaction. The swirled mixer was heated to a temperature of $250\,{}^\circ {\rm C}$ with the aid of a heating tape to prevent crystallization of ferrocene as it travels through the region. When the desired reactor temperature was reached, ferrocene was introduced into the vaporizer and heated until evaporation. Then, acetylene and argon were fed into the unit as the ferrocene evaporates and the electric motor at the reactor exit was switched on. The ferrocene evaporation was monitored through the glass window on the vaporizer. The reaction was stopped when all the ferrocene has evaporated from the vaporizer and then the reactor was allowed to cool down. Thereafter, the spun yarn was collected from the surface of the motor shaft. Experimental conditions under which the yarns were produced are shown in table 1.

Table 1. Experimental conditions for the production of the CNT yarn.

Sample	Reactor temperature $({}^\circ {\rm C})$	Acetylene flow rate (ml min−1)	Notation (Y-T-F)
1	900	100	Y-900-100
2	900	135	Y-900-135
3	1000	100	Y-1000-100
4	1000	135	Y-1000-135

Y-T-F: Y  =  yarn, T  =  temperature, F  =  flow rate.

SEM micrographs of the CNT yarns were obtained using an FEI Quanta 200 ESEM operated at 30 kV to check the morphology. A spatula tip of each sample was mounted onto an aluminum stub using carbon tapes. Two layers of gold-palladium coating, each of 10 nm in thickness, were applied to the mounted samples before taking images of the samples using the microscope.

FEI Tecnai T12 TEM operated at 120 kV was used to obtain 2D images of the CNT fibres. To prepare the samples for TEM analysis, a spatula tip of each of the samples was ultrasonically dissolved in 5 ml of methanol until they were well dispersed and formed a suspension in the solvent. A drop of this suspension was spread onto a copper grid and the copper grid was mounted onto the sample holder and loaded onto the instrument.

Raman spectra were obtained using a Horiba LabRAM HR Raman spectrometer equipped with an Olympus BX41 microscope. A Lexel argon ion laser at a wavelength of 514.5 nm (green line) was used to excite Raman scattering in the samples. A 100  ×  objective was used to collect the backscattered light, which was dispersed via a 600 lines mm⁻¹ grating onto a liquid nitrogen cooled charge coupled device (CCD) detector. The laser power at the samples was ~0.4 mW to prevent localized heating.

Four-probe method tests of the CNT yarn samples were conducted using a High Field Cryogen Free Measurement System (Cryogenic Ltd). Double-sided tape was used to mount a small quantity of each of the samples onto the sample holder. The mounted sample was pressed into a $5\times 5{\rm mm}$ mat. Silver conductive paint was applied to the four corners of the mat to provide contact between the mat and the sample holder. The paint was allowed to dry before the sample holder was mounted on a resistivity probe and connected to a four terminal configuration. Measurements were taken at room temperature by passing a voltage of 0–2 V through each sample. The electrical conductivity of each sample was calculated as described elsewhere [19] and presented below.

The gradient of the linear I-V curve obtained for each sample was used to estimate the resistance of each sample in accordance with Ohm's law, using equation (1):

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle R~\left( \Omega \right)=\frac{1}{{\rm gradient}}.\nonumber \end{align} \tag{ 1 }$$

The resistivity of each sample was calculated using equation (2):

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \rho ~\left( \Omega \cdot {\rm m} \right)=\frac{RA}{L}.\nonumber \end{align} \tag{ 2 }$$

The electrical conductivity of each sample was calculated using equation (3):

$$\begin{align} \newcommand{\e}{{\rm e}} \displaystyle \sigma \left( {\rm S }\,{{{\rm m}}^{-1}} \right)=\frac{1}{\rho }.\nonumber \end{align} \tag{ 3 }$$

Where $R$ is the resistance of the sample, $A$ is the cross-sectional area of the mat (equivalent to $25\,{\rm m}{{{\rm m}}^{2}}$), $L$ is the length of the mat (equivalent to 5 mm), $\rho$ is the resistivity of the sample and $\sigma$ is the electrical conductivity of the sample.

3.1.1. Morphology.

From the SEM micrographs shown in figure 2, it could be observed that there is presence of impurities in all the samples. These impurities are also evident in the TEM micrographs depicted in figure 3, where the nanomaterial is obscured with agglomeration of black particles. The agglomeration of black particles present could be attributed to the iron particle residues from the catalyst and perhaps, presence of amorphous carbon due to incomplete nucleation of carbon precursor [4]. This observation was confirmed by the elemental composition obtained from energy-dispersive x-ray spectroscopy (EDS) on sample Y-900-100, which showed the presence of copper and iron impurities in the sample (see figure 4). Other impurities obtained could be attributed to the presence of impurities in the ferrocene catalyst used in the synthesis.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The TEM images depicted in figure 3 reveal hollow tubular structure with inner diameter and length of several nanometers, thus confirming the presence of CNTs [6]. As shown in figure 3(a), the sample contained metal-filled CNTs with traces of iron impurities trapped within the hollow core of the CNTs. This could be attributed to the presence of excess iron catalyst particles which could not form nucleation sites for CNT growth [20]. Furthermore, figure 3(d) shows a knot-like structure which could be attributed to the spun of CNT bundles to form a twisted yarn structure [21].

Upon the analysis of the SEM images depicted in figure 2, formation of fibrous structures was observed in sample Y-900-135 (figure 2(b)) and sample Y-1000-135 (figure 2(d)). However, due to the large amount of catalyst residues present in sample Y-900-135, figure 2(b) does not provide a reliable indication that the structure resembles CNT yarn. Instead, an instrument with a higher resolution, such as the FEI Nova Nanolab 600 FEG-SEM/FIB, was required to assess the degree of alignment of CNTs within the structure before it can be deduced that the fibrous structure in sample Y-900-135 was indeed CNT yarn.

The SEM micrograph of sample Y-1000-135 (figure 2(d)) displays a fibrous structure that is continuous and resembles that of a yarn, such as cotton or wool. Additional SEM micrographs of the fibrous structure are presented in figure 5. Images shown in figure 5 provide evidence of a high degree of alignment parallel to the fibre direction and throughout the fibre thickness. Since CNTs are present in sample Y-1000-135 and the sample comprised mainly of carbon and iron as aforementioned, it can be deduced that the external morphology of the fibrous structure is a result of CNTs aligning into a fibre, thus indicating that CNT yarn was successfully spun.

Zoom In Zoom Out Reset image size

3.1.2. Raman spectroscopy of the as produced yarn.

The graphitization of the as-produced yarn samples was checked using Raman spectroscopy. A Raman spectrum of each sample is shown in figure 6. For each sample, Raman scattering was excited at three randomly selected spots in order to provide a better representation of the overall surface of the samples.

Zoom In Zoom Out Reset image size

In a typical Raman spectrum of CNTs, two characteristic peaks are observable: the D-band which is located between $1330\,{\rm c}{{{\rm m}}^{-1}}$ and $1360\,{\rm c}{{{\rm m}}^{-1}}$ and the G-band which is located between $1550\,{\rm c}{{{\rm m}}^{-1}}$ and $1600\,{\rm c}{{{\rm m}}^{-1}}$ [22, 23]. The D-band is assigned to disordered carbon or defects generated by the deformation of the original sp²-based structure. The G-band is representative of the degree of graphitization associated with the nanotube growth or corresponds to the stretching mode in the graphene plane [24]. As depicted in figure 6, the D-band and G-band are clearly identifiable within the aforementioned ranges in the Raman spectra of all the samples. This confirms that CNTs were produced and present in each sample. The noise observed in the spectra could be attributed to the iron impurities present in the samples [25].

The intensity ratio (D/G ratio) is an index of the degree of surface structural deformation, where a greater ratio signifies that greater disorder is present [26]. The obtained intensity ratio of each sample is listed in table 2. As shown in table 2, sample Y-900-100 is associated with the highest intensity ratio, indicating that it is more disordered than the other samples. However, sample Y-1000-135 possessed the lowest intensity ratio, indicating that it exhibits the highest degree of graphitization. The greater amount of disorder in sample Y-900-100 could be attributed to the presence of both carbon nanospheres and nanotubes in the sample, while sample Y-1000-135 exhibits more order as the CNTs have been aligned into fibrous structures.

Table 2. Intensity ratios of the as-produced CNT yarns.

Sample no	Sample	D/G ratio
1	Y-900-100	0.6657
2	Y-900-135	0.5202
3	Y-1000-100	0.5197
4	Y-1000-135	0.5096

For the same acetylene flow of 135 ml min⁻¹, it was observed that the higher reaction temperature produced a sample with less disorder (sample Y-1000-135). This result supports the findings of Iyuke and Simate [27], who reported that the crystallinity of synthesized CNTs is dependent on the pyrolysis temperature such that a higher temperature gives a better crystal structure.

From the four-probe method tests, typical linear ohmic behaviour was observed for all the as-synthesized samples, indicating that the samples were metallic in nature and electrically conductive. The ohmic behaviour, which describes the curfrent–voltage (I-V) characteristics of the samples, is depicted in figure 7. The average electrical conductivity of each sample is shown in table 3.

Zoom In Zoom Out Reset image size

The electrical conductivity of all samples is lower than that of typical CNT fibre reported in literature and lower than that of the conducting wires such as copper [28]. This could be attributed to the high impurity content of the yarn and structural flaws such as misalignment, local entanglements, poor densification or the presence of amorphous carbon [28]. At the same acetylene flow rate (135 ml min⁻¹) but different temperatures ($900\,{}^\circ {\rm C}$ and $1000\,{}^\circ {\rm C}$), sample Y-1000-135 exhibited a higher conductivity than sample Y-900-135, indicating that a higher reaction temperature results in a more conductive fibre, which has also been observed in literature. Niven et al [29] reported an increase in the electrical conductivity of yarn at increasing annealing temperature because resistance decreases with temperature. This is also in accordance with the results from the Raman spectroscopy and the SEM; where we observed that sample Y-1000-135 had a lower degree of disorder which is favourable to a more conductive fibre.

However, sample Y-1000-100 displayed a better conductivity than sample Y-1000-135, whereas sample Y-1000-135 has a lower degree of disorder according to Raman spectra. This inconsistency could be attributed to the difference in their chirality vectors. The electronic structure of CNTs does not depend on degree of disorder only, but also on the chirality vector of the CNTs [30]. Furthermore, Miao [21] reported the effect of yarn porosity on its conductivity measurement. Since the conductivity of the yarn was determined based on the cross-sectional area of the sample mat, there is possibility of fewer CNTs in a unit area and fewer CNT-to-CNT contact points in the mat cross-section could also affect the conductivity measurement.