Abstract
The structure of carbon materials can be controlled using a magnetic field, enhancing their functional properties. Most of the magnetic-field effects on carbon material growth were found to originate from the magnetic-field orientation. However, we observed that the magnetic-field orientation did not affect the growth of single-walled carbon nanotubes (SWCNTs); instead, under a magnetic field of 10 T, the preferential growth of metallic SWCNTs (1-nm diameter) was observed using chemical vapor deposition and liquid decomposition, suggesting chirality selectivity. Raman and X-ray photoelectron spectra showed that the defect structure and oxygen content of SWCNTs increased with increasing magnetic-field intensity. Therefore, thin metallic nanotubes can be selectively grown by applying a high magnetic field in environments where nanotubes are relatively difficult to form.
Introduction
Single-walled carbon nanotubes (SWCNTs) exhibit different electrical properties (metallic and semiconducting) depending on their chiral structure [1,2]. Recently, devices have been developed using these characteristics. Thus, controlling SWCNTs chirality is crucial for their applications. Currently, SWCNTs with uniform metallic and semiconducting properties and diameter distribution can be obtained through separation by electrophoresis [3,4] and gel chromatography [5–7]. Although it has been reported that physical properties can be directly controlled under limited conditions [8], purification mainly depends on separation [9].
Magnetic fields can externally control the carbon structure without contact. Carbon materials with the sp2 structure, which produce an extensive carbon hexagonal network of planes called graphene sheets, have a large anisotropic magnetic susceptibility because of the difference between the easy and difficult axes of magnetization caused by the carbon hexagonal network’s structure; thus, magnetic orientation occurs under a strong magnetic field [10–13]. In the case of carbon nanotubes, multi-walled carbon nanotubes (MWCNTs), which have many atoms dispersed in a solvent, are oriented under a strong magnetic field [14,15]. We have previously reported that when a magnetic field is applied to the preparation process of MWCNTs, nanotubes grow while oriented parallel to the magnetic flux direction [16]. Our approach which involved CNT synthesis being performed under a strong magnetic field of 10 T, at a level generated by a superconducting magnet, was unusual and rare.
We also investigated that the effects of a magnetic field during the growth of SWCNTs because SWCNTs were reported to be slightly oriented by electrical magnets [17]. Although no alignment was observed in our study, the number of metallic SWCNTs increased with the application of a magnetic field, which was confirmed by Raman scattering measurements. Although there have been cases in which the diameter of the nanotubes was limited for SWCNTs when a permanent magnet was placed near the arc discharge [18], there have been no cases in which the electrical properties have been controlled. To artificially place SWCNTs obtained by separation or generated on electrodes on a substrate would be a difficult task in creating devices. The most significant feature of our report was the selective growth of metallic SWCNTs directly on the substrate. To use metallic SWCNTs as a conductive material, a catalyst can be applied to the circuit and the SWCNTs will grow directly on it, which is of great value in terms of engineering applications. In this research, the details are investigated by changing the synthesis conditions and approach.
Experimental procedure
The synthesis of SWCNTs was conducted using two approaches [19]. In the first approach, CNTs were prepared using chemical vapor deposition (CVD) of ethanol on a Co/Mo-deposited quartz plate at 1073 K under the magnetic fields of up to 10 T using a superconducting magnet (HF-10-100 VHT-4, Sumitomo Heavy Industries, Ltd.) and a home-made furnace system (Schematic drawing of a SWCNTs preparation system by CVD shows in Supplementary Fig. S1) [13,16]. The flow rate of the saturated ethanol vapor containing the Ar gas was controlled between 500 and 2500 sccm. Quartz plates (5 × 10 × 1 mm3), preheated at 773 K in the air for 5 min, were dipped into a 0.01% ethanol solution of molybdenum (II) acetate dimer (Sigma-Aldrich) for 10 min and pulled up at a rate of 4 cm min−1, followed by heating at 673 K in the air for 5 min. In the same manner as in the Mo deposition, the Mo-deposited plate was dip-coated using a 0.002% ethanol solution of cobalt (II) acetate tetrahydrate (Nacalai Tesque Inc.). The plates were set at the center of the magnet, where the magnetic-field homogeneity was within 3%. In the second approach, liquid decomposition (LD) method was used [20,21]. Carbon paper (TGP-H-090, Toray Industries, Inc.), used as the carbon source and substrate, was immersed in a methanol solution of 1-mM cobalt (II) acetate tetrahydrate (as a catalyst source), and the carbon paper was resistively heated at 1053 K to vaporize methanol, after which carbon materials were grown on the carbon paper. The chamber with the synthesis system was placed in a superconducting magnet (Schematic drawing of a SWCNTs preparation system by LD shows in Supplementary Fig. S2). Vibration bands were measured employing a JASCO NRS-3100 microscopic Raman spectrophotometer using laser excitation at 532 and 785 nm. The morphology of the prepared carbon was analyzed using scanning electron microscopy (SEM; JEOL JSM-7600F) and transmission electron microscopy (TEM; JEOL JEM-2100F(HT)). The as-grown CNTs on carbon paper were observed using SEM. The chemical bonding states of the produced carbon materials were determined using X-ray photoelectron spectroscopy (XPS; JEOL JPS-9010TR). Mg Kα and Al Kα were used as X-ray sources.
Results and discussion
Figure 1(a) shows the radial breathing mode (RBM) and G- and D-band regions of the Raman spectra of SWCNTs grown on a quartz plate with one coating of a catalyst at an excitation wavelength of 532 nm under magnetic fields of up to 10 T. All spectra were normalized based on the most intense peak observed around 1580 cm−1 (the G-band peak). The Raman shift of the RBM (vRBM) is specific to SWCNTs and can be converted to the diameter of SWCNTs (d) using Eq. ((1)) [22]:
(a) Raman spectra of SWCNTs, excited at 532 nm prepared using the CVD approach under magnetic fields of up to 10 T. (b) Results of multiple measurements collected from the samples prepared under the same conditions. (c) Relation between I G ∕I D values and diameters of SWCNTs prepared under 0 (closed squares) and 10 T (open circles).
Figure 2(a) shows the Raman spectra of SWCNTs prepared using the LD approach, measured at an excitation wavelength of 532 nm. This result exhibits the same trend as that demonstrated for the SWCNTs prepared using the CVD approach (previous paragraph), indicating that this phenomenon is independent of the preparation approach. However, the same sample excited at 785 nm showed no change in the RBM, as illustrated in Fig. 2(b). In the case of excitation at 785 nm, the peaks observed at approximately 180 and 230 cm−1 corresponded to metallic and semiconducting SWCNTs, respectively [23,24]. The G- and D-band regions are also illustrated in each figure. Regardless of the excitation wavelength and the preparation method, the intensity of the D-band peak increased with increasing magnetic-field strength, indicating an overall decrease in crystallinity, which is the same as observed when using the CVD approach.
Raman spectra of SWCNTs excited at (a) 532 and (b) 785 nm, prepared using the LD approach under magnetic fields of up to 10 T.
Figure 3 shows the growth of SWCNTs and the magnetic-field effect when the catalyst, Co2+, were added at concentrations other than 1 mM. Although CNTs were synthesized with the catalyst concentrations up to 10 mM using the LD approach, SWCNT growth was observed only at concentrations of >1 mM. The magnetic-field effect was observed only at Co2+ concentration of 1 mM, as shown in Fig. 2. In the case of a Co2+ concentration of >5 mM, i.e., when SWCNTs were easier to form, relatively thin SWCNTs with a diameter around 0.9 nm were selectively grown, irrespective of the application of a magnetic field. At higher catalyst concentrations, it is reasonable to assume that SWCNTs grow more easily, indicating that SWCNTs with diameters around 1.0 nm are more likely to grow. Metallic SWCNTs with a 1.0-nm diameter may have formed because of low catalyst concentrations under the magnetic field.
Raman spectra of SWCNTs prepared using the LD approach via the thermal decomposition of a methanol solution containing cobalt (II) acetate tetrahydrate at various concentrations at 1053 K and 0 T (solid lines) or 10 T (dashed lines).
Another carbon material other than SWCNTs was formed at the same time when the amount of catalyst was small. Figure 4 (a) shows the SEM image of the carbon paper substrate surface, and Figs 4(b–d) show the TEM images of SWCNTs grown using a catalyst, MWCNTs, and carbon black, respectively. The RBM in the Raman spectra was obtained from only SWCNTs, while the G- and D-bands were obtained from all the carbon structures except SWCNTs. The morphologies of nanocarbons other than SWCNTs change when a magnetic field is applied [26,27]. Although we were unable to directly observe the change in the diameter of the product via TEM, this is largely owing to the spatial resolution limit in the present case.
Electron microscopy images of the carbon products prepared using the LD method with 1-mM Co+ in methanol at 0 T. (a) SEM image of the products grown on carbon paper prepared at 0 T. (b–d) TEM images of (b) SWCNTs; (c) MWCNTs; (d) carbon black.
XPS measurements were conducted to reveal structural changes in the prepared carbon materials. The X-ray photoelectron wide spectra of the carbon products prepared using the LD approach with 1-mM Co2+ in methanol showed C 1s, O 1s, and Co 2p 3/2 at approximately 284, 532, and 780 eV, respectively. Figure 5(a) shows the magnified view of the C 1s and O 1s peaks. The O 1s peak intensity tends to increase with respect to the C 1s peak intensity for the product obtained under 10 T. Figure 5(b) shows the XPS intensity ratios of the O 1s peak to the C 1s peak (I O ∕I C ). The O/C ratio tends to increase with increasing magnetic-field strength, showing that more oxygen was introduced into the structure of the product when the synthesis was conducted in a magnetic-field. This agrees with the decrease in the I G ∕I D value of Raman spectra with magnetic-field application (e.g., Fig. 1(c)), and the increase in the defect fraction can be ascribed to the increase in the oxygen content.
XPS spectra of carbon products prepared using the LD approach with Co2+ (1 mM) under 0-T magnetic field (solid lines) and 10-T magnetic field (dashed lines). (a) Wide spectra of around C 1s and O 1s regions. (b) I O /I C versus magnetic field.
The reaction mechanism of SWCNTs and the influence of the magnetic field on the mechanism are discussed herein. Catalyst particles are formed in the first step; then, the carbon source (methanol in this case) is thermally decomposed to provide a carbon source on the catalyst surface. The carbon atoms dissolve into the metal and form carbides. Subsequently, as shown in Fig. 6, when carbon atoms reach saturation precipitate on the surface, nucleation and the cap structure assembly occur and the nanotube grows [28]. Although amorphous carbon and other materials precipitate on the catalyst surface simultaneously, the growth of SWCNTs stops once the entire catalyst surface is covered with amorphous carbon. The increase in the carbide component under a magnetic field may inhibit the catalytic activity. The SWCNTs that are easy to grow demonstrate selective growth, even in an environment that is unfavorable to SWCNT growth.
SWCNT formation mechanisms in the absence and presence of the magnetic field.
We analyzed the carbon–catalyst bonding states to confirm this assumption. Figure 7 shows the X-ray photoelectron narrow spectra around the C 1s (Fig. 7(a)) and O 1s (Fig. 7(b)) regions, including peak deconvolution results. The results of peak area analysis (see Supplementary Table S1, S2) show an increase in the metal carbide and C–O functional group in the products [29]. The metal carbide peak observed around 778 eV was increased for the product prepared under the magnetic field, as shown in the Co 2p spectrum presented in Fig. 7(c). The dissolved oxygen is completely removed during the experiment, indicating that the oxygen is generated due to the decomposition of alcohol and the metal oxide added as a catalyst source. Therefore, the reactivity of the oxygen radicals formed via the decomposition of the raw alcohol in the vicinity of the metal catalyst may have increased under the magnetic field.
The magnetic field–dependent properties of structures involve extremely interesting phenomena and could offer a valuable option in developing functional materials and devices; therefore, we discuss the industrial applicability of selective growth of metallic SWCNTs on substrate. A 100-mm diameter bore of the superconducting magnet was used in the experiment of CVD method, and the effective space for synthesis was only 23 mm because of the cooling layer and the electric furnace. It may seem difficult to manufacture a product industrially using a superconducting magnet, but superconducting magnets are commercially available with bore diameters large enough to be industrial-scale production, such as a 40-cm diameter superconducting magnet that generates 7-T magnetic field for conducting MRI. The CVD method is a commonly used method for the synthesis of SWCNTs, and if the plant is properly placed inside the magnet, large-scale synthesis is viable. It is possible to leave only the SWCNTs on the substrate by burning off the impurities.
The application of a magnetic field facilitates the selective growth of SWCNTs. Regardless of the production approach, such as the CVD and LD, relatively thin diameters and metallic SWCNT compared to no magnetic field application were selectively obtained. However, the effect was not drastic and was observed only in limited environments, i.e., when the catalyst concentration was low. Although the proposed approach may not be practical for the mass production of SWCNTs, this technique may be used to grow a small amount of SWCNTs directly on a substrate for conductive device applications.
Footnotes
Acknowledgements
This research was partly supported by a Grant-in-Aid (No. 24655008) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and the Program for Fostering Regional Innovation in Nagano, established by MEXT, Japan.