A bimetallic cobalt-iron catalyst was utilized to demonstrate the growth of multiwalled carbon nanotubes (CNTs) at low gas pressure through thermal chemical vapor deposition. The characteristics of multiwalled CNTs were investigated based on the effects of catalyst thickness and gas pressure variation. The results revealed that the average diameter of nanotubes increased with increasing catalyst thickness, which can be correlated to the increase in particle size. The growth rate of the nanotubes also increased significantly by ~2.5 times with further increment of gas pressure from 0.5 Torr to 1.0 Torr. Rapid growth rate of nanotubes was observed at a catalyst thickness of 6 nm, but it decreased with the increase in catalyst thickness. The higher composition of 50% cobalt in the cobalt-iron catalyst showed improvement in the growth rate of nanotubes and the quality of nanotube structures compared with that of 20% cobalt. For the electrical properties, the measured sheet resistance decreased with the increase in the height of nanotubes because of higher growth rate. This behavior is likely due to the larger contact area of nanotubes, which improved electron hopping from one localized tube to another.

1. Introduction

Carbon nanotubes (CNTs) have elicited much interest due to their excellent electrical and mechanical properties that make them desirable for numerous potential applications such as field emission [1], atomic force microscopy tips [2], gas and vapor sensors [3], supercapacitors [4], and transistors [5]. To achieve maximum potential in target application, nanotubes with well-defined diameter, length, position, and optimized growth direction are necessary. When developing vertically aligned CNTs grown by chemical vapor deposition (CVD), the growth of the CNTs depends on the catalyst film thickness, substrate type, ratio of feed gas, temperature, and gas pressure. Several studies have been carried out to grow multiwalled carbon nanotubes (MWCNTs) from CVD or plasma-enhanced CVD using single-metal Co, Fe, and Ni as catalysts on various substrates [68]. To further improve the density of MWCNTs, plasma pretreatment is performed on these typical catalyst particles to enhance catalyst formation and stability onto the substrate [9]. Various studies on the effectiveness of using multilayered catalysts or alloys, such as Fe–Ni [10, 11], Fe–Mo [12], and Co–Mo [13], on CNTs have also been carried out. Combining transition metals as hybrid catalysts for the synthesis of nanotubes is also believed to result in higher yield and better quality [14]. Moreover, particle aggregation due to the addition of metal catalysts could be prevented, thereby improving the shape, size, and morphology of particles, and thus, significantly affecting the growth of CNTs.

In previous studies, thin film catalysts have been formed on various substrates by either sputtering or electron-beam evaporation [15, 16]. The catalyst film is then nucleated into particles by thermal annealing [17] or plasma treatment [18] to form the seed layer. Although the formation of catalyst particles is a key factor that determines the growth mechanism and nanotube diameter, only a few studies have explored the role of catalyst thickness in the growth process [19, 20], where strong correlation exists between the catalyst particle size and the growth of the nanotubes. The variation of nanotube structures can be formed during synthesis at different gas pressures [21]. Most researchers have concentrated on high gas pressure, typically from 1 Torr to 20 Torr, or atmospheric pressure to achieve high yield of CNTs [22, 23]. However, investigations on the viability of growing CNTs using hybrid metal catalysts at lower pressure are still lacking.

In this paper, a study on thermally grown MWCNTs using bimetallic cobalt-iron catalyst at low pressure is presented. The parameters affecting the growth of the nanotubes are investigated by varying the initial catalyst thickness and the gas pressure at low pressure. Results show that these parameters strongly influence nanotube diameter and growth rate. The results also provide a correlation between the formation of nanotube array and the measured sheet resistance.

2. Experiment Details

The synthesis of CNTs was performed by thermal CVD. Prior to the growth process, a thin film of bimetallic Co–Fe was deposited as catalyst onto a 10 nm TiN barrier layer/thermally oxidized Si (100) substrate through radio frequency-magnetron sputtering at a base pressure of ~10−3 mbar. The Co–Fe composition varied at 20% and 50% with respect to %Co, whereas the catalyst thickness varied at 4, 6, 8, and 10 nm. A short nucleation step involving plasma pretreatment of Co–Fe catalyst layer was performed for 10 min in the presence of hydrogen at 700°C with 200 W plasma power to break up the films into small particles and reduce the impurities. The nanotubes were then grown on the catalyst particles with an acetylene flow rate of 50 sccm for 5 min by varying the gas pressure at 0.5, 0.75, and 1.0 Torr. Both the nucleated catalyst particles and the grown CNTs were characterized using a JEOL field emission scanning electron microscopy system. Raman spectra were obtained on the nanotube samples with a NT-MDT Raman spectroscopy system using a 473 nm air-cooled blue laser as an excitation for analysis. Emission intensities from the samples were measured from the peak height. For the electrical properties, the sheet resistance of the samples was measured at room temperature with an Ecopia-HMS5300 measurement system based on the van der Pauw method under a current excitation of 1 mA.

3. Results and Discussion

3.1. Catalyst Particle Formation

Figure 1(a) illustrates the SEM images of the bimetallic Co–Fe catalyst distributed in the form of nanoparticles at 20% and 50% with respect to %Co after plasma pretreatment at 700°C. The Co–Fe particles, which acted as a seeding layer prior to the growth of nanotubes, were formed during heating due to the surface tension and compressive stress between the TiN/Si (100) substrate and the Co–Fe catalyst as a result of the lattice mismatch in thermal expansion coefficients [24, 25]. Based on the results, the size distribution of Co–Fe particles depends on the initial catalyst thickness of the sputtered Co–Fe, which in turn determines the particle formation. Figure 1(b) shows the dependence of average particle size and particle density on initial catalyst thickness. Catalyst formation at a 4 nm thick layer for both catalyst compositions was the densest, with small homogeneous particles due to small coalescence of the particles occurring during agglomeration. For a short nucleation time, a thinner catalyst was found to rapidly agglomerate, thereby forming small homogeneous particle sizes with minimal interparticle distance. Meanwhile, a large-sized distribution with cluster-like particles possessing lower density was observed at the thicker Co–Fe catalyst. This occurrence is attributed to the nucleation and agglomeration of the particles that were incomplete in excess of the thick catalyst, resulting in broader size distribution.

In terms of Co composition in the Co–Fe catalyst, the Co50Fe50 catalyst exhibited higher particle density by more than ~40% at the 4 nm thick catalyst due to the formation of smaller particle sizes compared with the Co20Fe80 catalyst. The particle density decreased with the increase in particle size because of the thicker catalyst. Therefore, a strong correlation possibly exists between the average particle size and the particle density with 4 nm to 10 nm thick Co–Fe catalysts.

3.2. Carbon Nanotube Growth

The influences of gas pressure during CNT growth are shown by the SEM images in Figure 2(a). No growth was observed at 0.5 Torr gas pressure on the 4 nm thick catalyst. However, highly dense “spaghetti-like” CNT structures were observed when the pressure was increased to 1.0 Torr. The growth behavior as a function of gas pressure can be explained by the reaction between the catalyst activity and the dissolution, diffusion, and precipitation rates of the carbon atoms [21]. At 0.5 Torr, the dissolution and diffusion of carbon atoms into the seeding layers are limited because of the low concentration of carbon atoms. The high surface-to-volume ratio of small particles is very sensitive to passivation at high temperature without plasma with thinner Co–Fe catalyst layers [26]. This condition allows the amorphous carbon to react with the catalyst, thereby preventing the carbon feedstocks from diffusing into the particles [27]. Due to limited carbon feedstocks during a short growth time, nanotube growth is likely to be suppressed, subsequently resulting in the formation of a carbon shell around the Co–Fe particles. Meanwhile, the high concentration of carbon atoms at higher pressure of 1.0 Torr increases the diffusion and precipitation rates of carbon atoms, thus enabling their dissolution into the seed layers at higher rate, diffusion through the seed layers, and initiation of the growth process to form CNTs.

Figure 2(b) shows the Raman analysis for CNTs grown with Co20Fe80 and Co50Fe50 catalysts. Two major peaks are found, namely, a -band at around ~1590 cm−1 and a -band at around ~1360 cm−1. The two-phonon -band (2D) appears at ~2690 cm−1. The results show that the present growth process only forms MWCNTs due to the absence of peaks at radial breathing mode frequency for single-walled carbon nanotubes. The -band represents the ordered graphite corresponding to the stretching mode of the C–C bonds in the graphite plane, whereas the -band represents the disordered graphite associated with defects and amorphous carbon. The crystalline quality of the nanotubes can be evaluated based on the ratio of peak intensities / (Figure 2(b)), except for the growth at a 0.5 Torr gas pressure, because no significant peak can be observed due to the lack of nanotube growth. The / increased with increasing gas pressure from 0.75 Torr to 1.0 Torr, which in turn reduces the defects in the nanotubes. Meanwhile, an increase in / up to ~8% was also observed when the composition of Co in the catalyst was varied from 20% to 50%. The amorphous carbons that may exist in thermally grown CNTs are not detectable through the SEM images. Alternatively, the observed increase in / may be an indicator of the reduction in the amount of amorphous carbon. The Raman spectra analysis revealed that the CNTs deposited on the TiN barrier layer are of MWCNTs, resulting from a base growth due to strong Co–Fe–TiN interactions.

The correlation between the average particle size and the average nanotube diameter is shown in Figure 3(a). The results indicate that the nanotube diameter is dependent on the particle size, and the trend shows a nonlinear behavior. The increased particle size resulting from thicker catalyst increases the diameter of the nanotubes during the growth process. By varying the gas pressure from 0.5 Torr to 1.0 Torr, the diameter of the nanotubes tends to increase by more than 30% and 14% for the Co20Fe80 and Co50Fe50 catalysts, respectively. The nanotube diameter is less than the particle size, which presents a similar behavior as described in the previous studies of Kukovitsky et al. and Nasibulin et al. [28, 29]. Nanotube formation may be restricted during vertical growth on the Co–Fe particles due to the strong interaction between the Co–Fe particles and the TiN layer. Therefore, the relationship between the particle size and the nanotube diameter becomes more complicated because gas pressure also influences nanotube diameter.

The growth rate curves which correspond to the catalyst composition and gas pressure as a function of initial catalyst thickness are demonstrated in Figure 3(b). The growth rate is obtained by measuring the height of the CNTs grown for 5 min. The growth rate of CNTs grown using both Co20Fe80 and Co50Fe50 catalysts increased when the gas pressure increased from 0.5 Torr to 1.0 Torr due to higher precipitation rate of carbon atoms. The growth rate curve increased first with increased catalyst thickness and then decreased when the catalyst thickness was more than 6 nm as a result of the increased particle size. Larger particle sizes are expected to be less energetic due to lower surface-to-volume ratio compared with smaller particle size. The diffusion and saturation of carbon atoms into the larger particles occurred at a lower rate, thereby forming short nanotubes at the end of the growth process. Based on the results, a significant growth rate of nanotubes was found on a 6 nm thick Co50Fe50 catalyst which was about ~1.4 times higher compared with that of the Co20Fe80 catalyst. Clearly, higher Co composition in the catalyst has a role in improving the CNT growth rate, as shown in Figures 4(a) and 4(b). The result is also attributed to the formation of smaller Co50Fe50 catalyst particles compared with that of Co20Fe80 (as shown in Figure 1(a)), which has greater supersaturation rate of carbon feedstock to enhance CNTs growth.

Figure 5(a) shows the sheet resistance graph of MWCNTs against the initial catalyst thickness. For all cases, the sheet resistance initially decreased and then increased when the catalyst thickness is more than 6 nm. The trend is opposite the results in Figure 3(b), so relating the contact resistance with the height of the nanotube is possible. Moreover, a much lower sheet resistance curve could also be seen on the nanotubes grown at higher gas pressure. For the 0.5 Torr gas pressure, the suppressed nanotubes grown on 4 nm thick catalysts of Co20Fe80 and Co50Fe50 in the form of carbon shells around the catalysts showed higher sheet resistance of about 17.4 and 13.8 kΩ/square, respectively. The lowest sheet resistance was observed in MWCNTs grown on 6 nm Co50Fe50 catalyst thickness at 1.0 Torr, where it decreased by about 98% relative to the highest sheet resistance. The formation of longer nanotubes grown on the 6 nm thick catalyst showed improved vertical alignment with higher density compared with those grown on the 4 nm thick catalyst, as shown in Figures 5(b) and 5(c). The presence of longer MWCNTs is expected to result in fewer end connections and larger contact area between the neighboring nanotubes due to the smallest gap among themselves, resulting in lower sheet resistance. This result suggests that electron transport in the nanotubes is limited by the tunneling effect between the gap and the adjacent nanotubes. The higher density of nanotubes resulting from longer nanotubes would increase the contact area among themselves. This condition improves electron hopping from one localized tube to another, thereby reducing resistance through the MWCNTs.

4. Conclusions

This paper presented an investigation on the effect of catalyst thickness and gas pressure on MWCNTs grown by CVD using bimetallic Co–Fe layer. The Co–Fe catalyst particle size (initial thickness) essentially influenced the nanotube diameter and the growth rate. The nanotube diameter was slightly lesser compared with the catalyst particle size. Thus, 1 : 1 ratio between the catalyst particle size and the nanotube diameter did not exist. Results also showed that nanotubes grew at a rapid rate when a 6 nm thick catalyst was used, and the rate appeared to be slower when the catalyst thickness was further increased to 10 nm. However, the growth rate of the nanotubes could be improved by increasing gas pressure during the growth process due to the higher concentration of the carbon feedstock. A significant growth rate of nanotubes appeared at 6 nm Co50Fe50 catalyst thickness, which was ~1.4 times higher compared with that of the Co20Fe80 catalyst due to the formation of smaller-sized particles. As a result, an improved electrical sheet resistance could be observed from the longer MWCNTs grown on the Co50Fe50 catalyst. The higher density of the MWCNTs from longer nanotube formation improved the contact area between the neighboring nanotubes, resulting in lower sheet resistance. Moreover, the Raman spectra results also showed that the quality of the CNTs grown on Co50Fe50 catalyst was better by about ~8% compared with that of the Co20Fe80 catalyst.


This research was supported by the National Nanotechnology Directorate Funding NND/ND/(2)/TD11-012 and the eScience Funding 01-03-04-SF0027 under the Ministry of Science, Technology, and Innovation (MOSTI), Malaysia, and ERGS 203/PMEKANIK/6730069 under the Ministry of Higher Education (MOHE), Malaysia.