Selective self-assembly of single walled carbon nanotubes in long steel tubing for chemical separations

Abstract

This paper reports the scaled-up self-assembly of single walled carbon nanotubes (SWCNTs) on the inside wall of a long silica-lined steel capillary tubing for applications such as chemical processing and separations. A unique one step self-assembly process has been developed and the effect of the substrate on the morphology of the CVD growth has been studied. With the aerosol spray of an ethanolic solution consisting of dissolved cobalt and molybdenum as metal catalysts and co-catalysts respectively, the catalyst was generated and activated in situ inside the interior of the tubing, in parallel with the synthesis of SWCNTs, thus eliminating the need to coat the substrate with the catalyst prior to the synthesis of the nanotubes. The presence of a silica layer on the steel tubing was found to be critical for the formation of SWCNTs. Gas chromatographic separation of aromatic compounds is demonstrated on the capillary tube.

---

Introduction

Ever since their discovery,¹ single walled carbon nanotubes (SWCNTs) have attracted the attention of the scientific community since they possess unique physical, chemical, and, electronic properties.^2–4 SWCNTs are generally synthesized by arc discharge,⁵ laser ablation⁶ or chemical vapor deposition (CVD). However, CVD is the method of choice when it comes to the self-assembly of CNTs over micro or macro structures.^7–10 Typical catalytic CVD involves the thermal pyrolysis of a precursor in the presence of transition metal nanoparticles, such as, Fe, Ni, and Co, which are known to act as catalysts⁸ in the synthesis of SWCNTs. An important requirement for the growth of SWCNTs is the nano-structuring of the catalyst because the diameter of the CNTs is commensurate with the diameters of the catalyst nanoparticles.¹¹ A wide range of precursors such as CO, hydrocarbons, alcohols, and ketones have been used to grow SWCNTs. Alcohols have shown some selectivity towards SWCNT formation and it has been hypothesized¹² that the OH radical formed at high temperatures oxidizes and therefore etches away the amorphous carbon impurity phase formed during nanotube growth.

It has been shown that substrate characteristics,¹³ such as porosity, crystallinity, and surface roughness, in addition to the choice of precursor, CVD conditions, and the catalyst influence the growth of single and multiple walled nanotubes. Optimizing so many variables is a challenging task, and limited understanding of these effects has undermined efforts to fabricate larger scale real world nanotube-based devices. However, for chromatographic applications, which would benefit from the high surface area and nanoscale adsorption properties of SWCNTs, large scale self-assembly of the type investigated here would be required.

SWCNT synthesis via CVD is typically a two-step process. The first step involves the coating of the substrate (typically Si or quartz) with a catalyst, or the metal impregnation of a catalyst support, e.g. silica, MgO, zeolites, and aerogels.⁸ Subsequent heat treatment leads to the formation of metal nano-particles. Catalyst coatings have been deposited by spin coating,¹⁴ dip coating,¹⁵ sputtering,¹⁶ and sol–gel¹⁷ techniques, or by merely placing the supported catalyst particles in a ceramic or quartz boat. Ago et al.¹⁸ have reported gas phase catalytic synthesis of SWCNTs using a reverse micelle solution containing Co–Mo nanoparticles. After catalyst deposition, the CVD synthesis of CNTs is carried out; thus involving two distinct processes. However, from the stand-point of process scale up, a single step process that combines catalyst deposition/preparation and CVD synthesis is preferred. Single step processes have, however, been less commonly used. One such method is the floating catalyst¹⁹ based CVD method, where a volatile organometallic precursor such as ferrocene is typically used to generate the catalyst particles during SWCNT growth.

Self-assembly of SWCNTs has been limited to substrates such as silicon, quartz, aerogels of silica or alumina of relatively small dimensions mainly used as field emitting devices,²⁰ tips for scanning probe microscopy (SPM),²¹ and as gas sensing devices.²² Self-assembly of MWCNTs has been reported on a stainless steel mesh.²³ Recently our group has reported the self-assembly of MWCNTs inside steel tubings for preconcentration⁹ and gas chromatography.²⁴ The scaling up of the self-assembly process on larger structures, such as a long metal capillary tubing, is relatively more complex, especially for SWCNTs. Several challenges face such an operation; these include the deposition of an uniform nano-sized catalyst, controlling the kinetics of the growth process, and selectively growing SWCNTs while avoiding the growth of MWCNTs and amorphous carbon. Also, SWCNT synthesis is a substrate specific process,¹³ which means the methods need to be tested and optimized if substrates are different. High pressure carbon monoxide (HiPCO)²⁵ and floating catalyst based CVD methods, which have succeeded in selectively synthesizing SWCNTs require temperatures above 1000 °C which are likely to melt metal substrates. This paper reports the development of a one-step process for the self-assembly of SWCNTs in a long capillary tube, which can be then used for chromatographic separation applications. It involves the aerosol spraying of an ethanolic solution (where ethanol is the carbon precursor) containing cobalt and molybdenum salts as metal catalyst and co-catalyst precursors, respectively. The catalyst and co-catalyst metal particles are generated and activated in situ, along with the synthesis of the SWCNTs. Thus, the separate coating of the substrate with catalyst prior to SWCNT synthesis is not required. The entire operation takes about 15 minutes for completion, is easy to scale up, and is economical.

Experimental

Cobalt nitrate hexahydrate, Co(NO₃)₂·6H₂O, and molybdenum acetate, (CH₃COOH)₂Mo, were dissolved in ethanol at 0.2 wt% and 0.05 wt% concentrations respectively. The dissolution process was assisted by sonication. The CVD system is shown in Fig. 1. The ethanol solution containing the dissolved catalyst acetate precursors was injected into the capillary metal tubing using a HPLC pump (Waters, Model 501). The typical flow rate of the solution was 100 µl min⁻¹. Hydrogen was simultaneously introduced into the steel tubing through a three way connector. The typical hydrogen flow rate was 40 cm³ min⁻¹. Check valves (R. S. Crum & Co., Mountainside, NJ) were placed on both lines to restrict back-flow. The CVD was performed typically for about 10 min. Since the substrate plays an important role, different tubing materials including 304 and 316 stainless steel (Alltech, Deerfield, IL) and silica lined tubings, such as Silcosteel™ and Sulfinert™ (Restek, Bellefonte, PA) were tested. The tubes were 1 meter long, with a 0.53 mm ID. Prior to CVD, the tubes were washed with ethanol and allowed to dry.

Fig. 1 Setup of the vapor phase catalytic synthesis of SWCNTs. The inset shows the inside of the steel tubing.

To study the CNT formation, 1 cm long segments were cut from the steel tube at five equidistant locations. The samples were cut open to expose the inside surface, and were analyzed by field emission-scanning electron microscopy. The distribution, surface coverage, and the thickness of the SWCNT coating were studied based on the SEM images. The presence of SWCNTs was confirmed by Raman spectroscopy performed at 632.8 nm excitation using a Jobin-Yvon/Horiba confocal micro-Raman system.

Results and discussion

Self-assembly of SWCNTs

Nanostructured iron was generated in situ on the steel tubing by oxidation with O₂ followed by reduction with H₂. This process has been described in detail in previous publications.^10,24 However the process generated only MWCNTs, no SWCNTs could be detected by Raman measurements. While ethylene as carbon precursor resulted in excellent MWCNT growth and good surface coverage, similar growth of SWCNTs was not achieved with ethanol, which showed large amounts of amorphous carbon as shown in Fig. 2a. In the absence of surface conditioning, the nanotubes were interspersed with larger amounts of amorphous carbon.

Fig. 2 SEM images: a) CVD morphology (30 min) with Fe inside bulk steel (304) as catalyst, b) typical 10 min CVD growth on 304 steel surface with Co and Mo catalysts, c) typical 10 min CVD growth on 316 steel surface with Co and Mo catalysts.

Co and Mo salts dissolved in ethanol were aerosolized in the presence of H₂ in the steel tubing. It was expected that nano-scale Co and Mo would result in SWCNT formation. When CNT formation inside the tube was observed by SEM, some nanotubes had small enough diameters (7–10 nm) that SWCNT formation was suspected. However, none could be detected by Raman spectroscopy. The bulk metal (iron) appeared to provide the dominant catalytic activity leading to MWCNTs and amorphous carbon formation. 304 stainless steel tube resulted in only patches of MWCNTs after 10 min CVD as shown in Fig. 2b. However, after 45 min CVD the entire tube was covered with CNTs with diameters ranging from 10–30 nm, interspersed with large amounts of amorphous carbon. On the other hand, 316 steel tubing exhibited profuse surface coverage of MWCNTs and amorphous carbon even after 10 minutes of CVD as shown in Fig. 2c. The diameters of the nanotubes varied widely from 20–120 nm. The difference in MWCNT formation and morphology on these two types of steel tubings may be attributed to the differences in their chemical compositions, their physical characteristics and their grain boundaries. Type 316 stainless steel is known to contain higher concentrations of molybdenum and nickel and lower concentrations of chromium than 304 steel. Larger diameters and growth density of the CNTs in 316 stainless steel may also be due to the higher rate of diffusion of carbon into its soft body.

The strategy for selective SWCNT growth required the prevention of iron from catalyzing MWCNT growth. Although the HiPCO process, which is carried out at temperatures above 1000 °C, uses iron as catalyst to form SWCNTs, it is possible that at 725 °C iron results in the formation of MWCNTs.⁸ Consequently silica lined stainless steel tubings, such as Silcosteel™ and Sulfinert™, were selected. These tubular substrates contained a 1.0–1.4 µm thick amorphous silica coating on the metal surface, which would provide a barrier to interaction of the carbon with iron during the CVD process. A catalyst for the SWCNT growth was therefore needed and was dispersed uniformly along the length of the tubing. The silica lined surface after and before catalyst deposition is shown in Fig. 3a and b.

Fig. 3 SEM images showing a) the dispersed catalyst layer on the steel surface; the inset shows a magnified view of the catalyst, and b) the surface of the steel without any catalyst.

The SEM images showed that during high temperature CVD, the silica coating developed small cracks, thus increasing the surface area and roughness as shown in Fig. 4a. These appear to provide sites for catalyst deposition. The nano-structured metal catalyst was generated in situ during the CVD process. The metal salt catalyst precursors were dissolved in ethanol. The pumped solution in conjunction with flowing hydrogen created an ethanol–catalyst precursor aerosol inside the tube. At high temperature (725 °C), the solution vaporized and distributed the catalyst along the whole length. The Co-nitrate precursor broke down to form Co nano-particles that were activated in the reducing H₂ environment. The metal then catalyzed SWCNT growth with ethanol serving as the carbon source. In order to study the morphology of the catalyst particles formed on the surface of the substrates, an aqueous catalyst solution was sprayed into the steel tubing in the presence of argon, while the other CVD conditions remained unchanged. It was difficult to see the dispersed catalyst on the rough surface of the silica-lined steel tube, consequently the steel tube was used in this study. Fig. 3a shows an SEM image of the catalyst deposition on the steel tube substrate. The images indicate that the catalyst layer was finely dispersed with a typical particle size less than 10 nm, with some as large as 15 nm. The SEM images indicated that the latter half of the tubing had a slightly denser catalyst layer. An SEM image of the steel surface pretreated at the same temperature with only water (without the catalyst) is also presented in Fig. 3b which provides a baseline for studying catalyst deposition.

Fig. 4 SEM images showing a) the rough surface of the Silcosteel™, b) the SWNT film upon the silica lined steel substrate, and c) the randomly aligned SWNTs.

The SEM images of the silica lined tube sections along the length of the tube revealed a randomly distributed layer of thin SWCNTs on the inside wall as shown in Fig. 4b and c. The surface coverage along the length varied, with the mid-section having a higher density of nanotubes and a thicker film. This trend could be due to the ends of the tube being relatively cooler than the mid-sections or due to the kinetics of nanotube formation. The variation in film thickness is presented in Table 1.

Table 1 Thickness (µm) distribution of the SWCNT coating measured along the 1 m long Sulfinert capillary metal tubing subjected to uniform temperatures

Tubing length/cm	SWCNT coating thickness/μm
2.5	0.025–0.05
12.5	0.2–0.3
25	0.2–0.4
37.5	0.2–0.4
50	0.2–0.4
62.5	0.2–0.3
75	0.025–0.05

---

Characterization of SWCNTs

The catalytic metal particles on the substrate and on the SWCNTs were detected using energy dispersive X-ray (EDX) analysis operated at 7 kV and at a scan time of 1000 s. The EDX profile as shown in Fig. 5a indicates the presence of cobalt and molybdenum catalysts. The signal intensity for molybdenum was significantly lower than for cobalt in line with their concentrations in the solution. The presence of SWCNTs was confirmed by Raman spectroscopy on all the sections of the tubing. Multiple tubes were analyzed to check reproducibility. The Raman spectrum is shown in Fig. 5, it showed the presence of the radial breathing mode (RBM) which is a characteristic of the SWCNTs. Based on the characteristic peaks at wavenumbers of 190, 217, 221, 248 and 287 cm⁻¹, the SWCNT diameters were computed from the following relation:²⁶ω_RBM = A/d + B, where A is 214.4 ± 2 cm⁻¹ and B is 18.7 ± 2 cm⁻¹, d is the diameter and ω_RBM is the Raman shift. The diameters of the SWCNTs were found to be between 0.8 and 1.25 ± 0.2 nm. At higher wavenumbers, all spectra showed two Raman bands at ∼1320 cm⁻¹ (D band) and ∼1590 cm⁻¹ (G band). The ratio of intensities of the G band to the D band over various sections along the length of the silica lined tubing varied from 1.85 to 0.75 suggesting the presence of CNTs with higher degrees of crystallinity in some areas, and higher degrees of disorder and defects at a few other areas. This heterogeneity in crystallinity could be a result of the self-assembly process via CVD. At a scaled up level (over a meter length), in such a tubular flow system, there is always an interplay of myriad mechanisms and free radicals in the formation of the nanotubes. This has been reported by our group in a recent publication.²⁷ Certain mechanisms predominate over others at different locations at different residence times. This eventually leads to the formation of CNTs with disparity over the length of the long tube.

Fig. 5 a) EDX profile of the SWCNT coating inside the capillary tubing. The profile is marked with the important peaks which are expected and correspond to C(K), O(K), Si(K), Co(L), Mo(L) peaks respectively. b) The Raman spectrum showing the RBM signals characteristic of the SWNT growth. c) The Raman spectrum showing the D and G signals.

Gas chromatography on SWCNTs

Gas chromatographic separations of a variety of analytes, such as alkanes, poly aromatic hydrocarbons (PAHs), alcohols, hydrocarbons, ketones and alkane isomers, were successfully carried out on the steel capillary tubing containing the self-assembled SWCNTs. A sample chromatogram of aromatic analytes is presented in Fig. 6 as a testimony to prove the concept of separation. The rest of the chromatograms are not presented here for brevity. Typical reproducibility in retention time measured as relative standard deviation (RSD) was less than 2%, which is comparable to those from commercial GC columns. The ability of the SWNT phase in the separation of analytes with a wide range of boiling points and volatility is quite exciting from the standpoint of chemical processing and separations. This was possible due to the stability of the SWCNT phase at high temperatures.

Fig. 6 Typical chromatogram generated from the SWCNT column showing the separation of aromatics. Conditions: 120 °C for 0.1 min, 45 °C min⁻¹ to 300 °C min⁻¹, flow rate of carrier gas 5.7 ml min⁻¹.

Conclusions

SWCNTs were self-assembled on the inside wall of a long silica coated stainless steel capillary tubing using a single step catalytic CVD process that is economical in time, effort and cost. To the best of our knowledge it is also the first time that SWCNT self-assembly has been reported at such a scaled up level inside a capillary tube. The silica lining was critical to the formation of SWCNTs, while uncoated, plain stainless steel tubings mainly formed MWCNTs. The use of these tubular substrates in the growth of SWCNTs represents a novel idea because it demonstrates the use of traditional metal platforms for fabricating SWCNT based devices for a real-world application, such as gas chromatographic separation using carbon nanotubes.

Acknowledgements

This research was supported by a grant from the US EPA STAR grant RD 830901.

References

1. S. Ijima and T. Ichihashi, Nature, 1993, 363, 603.
2. H. Dai, Acc. Chem. Res., 2002, 35, 1035.
3. A. Peigney, Nat. Mater., 2003, 2, 15.
4. J. Kong, N. R. Franklin, C. Zhou, M. G. Chapline, S. Peng, K. Cho and H. Dai, Science, 2000, 287, 1801.
5. C. Journet, W. K. Maser, P. Bernier, A. Loiseau, M. Lamy de la Chappelle and S. Lefrant, Nature, 1997, 388, 756.
6. T. Guo, P. Nikolaev, A. Thess, D. T. Colbert and R. E. Smalley, Chem. Phys. Lett., 1995, 243, 49.
7. C. P. Deck and K. Vecchio, Carbon, 2005, 43, 2608.
8. A.-C. Dupuis, Prog. Mater. Sci., 2005, 50, 929.
9. C. Saridara, R. Brukh, Z. Iqbal and S. Mitra, Anal. Chem., 2005, 77, 1183.
10. M. Karwa, Z. Iqbal and S. Mitra, Carbon, 2005, 44, 1235.
11. C. L. Cheung, A. Kurtz, H. Park and C. M. Lieber, J. Phys. Chem. B, 2002, 106, 2429.
12. S. Maruyama, R. Kojima, Y. Miyauchi, S. Chiashi and M. Kohno, Chem. Phys. Lett., 2002, 360, 229.
13. J. W. Ward, B. Q. Wei and P. M. Ajayan, Chem. Phys. Lett., 2003, 376, 717.
14. N. R. Franklin, Y. Li, R. J. Chen, A. Javey and H. Dai, Appl. Phys. Lett., 2001, 79, 4571.
15. Y. Murakami, Y. Miyauchi, S. Chiashi and S. Maruyama, Chem. Phys. Lett., 2003, 377, 49.
16. L. Delzeit, B. Chen, A. Cassell, R. Stevens, C. Nguyen and M. Meyyappan, Chem. Phys. Lett., 2001, 348, 368.
17. D. Méhn, A. Fonseca, G. Bister and J. B. Nagy, Chem. Phys. Lett., 2004, 393, 378.
18. H. Ago, S. Ohshima, K. Tsukuagoshi, M. Tsuji and M. Yumura, Curr. Appl. Phys., 2005, 5, 128.
19. S. Bai, F. Li, Q. Yang, H.-M. Cheng and J. Bai, Chem. Phys. Lett., 2003, 376, 83.
20. J. M. Bonard, H. Kind, T. Stockli and L. A. Nilsson, Solid-State Electron., 2001, 45, 893.
21. S. S. Wong, E. Joselevich, A. T. Woodley, C. L. Cheung and C. M. Lieber, Nature, 1998, 394, 52.
22. J. Li, Y. Lu, Q. Ye, M. Cinke, J. Han and M. Meyyappan, Nano Lett., 2003, 3, 929.
23. R. L. V. Wal and L. J. Hall, Carbon, 2003, 41, 659.
24. C. Saridara and S. Mitra, Anal. Chem., 2005, 77, 7094.
25. W. Zhou, Y. H. Ooi, R. Russo, P. Papanek, D. E. Luzzi, J. E. Fischer, M. J. Bronikowski, P. A. Willis and R. E. Smalley, Chem. Phys. Lett., 2001, 350, 6.
26. H. Telg, J. Maultzsch, S. Reich, F. Hennrich and C. Thomsen, Phys. Rev. Lett., 2004, 93, 177401.
27. R. Brukh and S. Mitra, Chem. Phys. Lett., 2006 , in press.

---