SEARCH

SEARCH BY CITATION

Keywords:

  • aluminum;
  • carbon nanotubes;
  • electron energy-loss spectroscopy;
  • electron microscopy;
  • iron

The discovery by Hata et al.1 of “super-long” carbon nanotubes (CNTs) grown by catalyst-driven, water-assisted chemical vapor deposition represented a major synthetic breakthrough for the integration of CNTs into future device architectures, such as chemical and physical sensors, heterogeneous catalyst arrays, or 3D micro- and nanoelectromechanical systems (MEMS and NEMS).2 Their work was preceded by the development of the HiPCO process by Smalley et al., which allowed, for the first time, a massive formation of unordered single-walled CNTs (SWCNTs).3 This discovery was followed shortly thereafter by the first growth process of long (up to 4 μm) vertically aligned SWCNTs by Muruyama et al.4 Recently, Hata et al. found that CO2, ethers, ketones, and alcohols are, in the same way as water, advantageous for the growth of arranged “super-long” CNTs.5 All growth enhancers contain oxygen atoms, and act as weak oxidants,4 which are able to etch away a growing carbon layer, which deactivates the active catalyst, and thus being detrimental to the growth of “super-long” CNTs.1, 6 The composition of the active catalyst in the “super-long” growth of CNTs needs further investigation, despite experimental efforts in the field of in situ characterization of CNT growth.7

Herein we report our findings regarding the nature, role, and mechanism of the catalyst composition in the growth of “super-long” CNTs. Our studies allow an understanding of the morphology and structure of the catalyst nanoparticles by high-resolution scanning transmission electron microscopy (STEM) tomograms and of their chemical composition by a combination of spectroscopic and diffraction techniques.

Besides catalyst dispersion, the properties of the substrate surface (roughness, active area, and electron-transfer ability to the catalyst surface) are critical for a high-yielding efficient CNT growth.812 Aluminum is known to work as an efficient buffer layer between catalyst and substrate under “super-long” CNT growth conditions. CNT growth starts with deposition of iron metal particles on a thin aluminum metal film deposited onto a Si wafer having a thin native SiO2 layer.1 We have chosen two different procedures for catalyst sample preparation, which allow direct observation of the catalyst’s structure and morphology and characterization of its chemical composition. The first set of experiments was carried out on transmission electron microscopy (TEM) grids and the second on a Si wafer substrate. In both cases we deposited a 10 nm-thick aluminum film, followed by a 1 nm-thick layer of iron metal. These deposited films were then heated directly to 750 °C. The resulting catalysts were then tested and found to be active in the water assisted chemical vapor deposition (CVD) growth of CNTs in an independent set of experiments (for details and characterization of the CNTs, see the Supporting Information).13 The chemical composition of the catalyst system was further studied to clarify whether (a) FexAly intermetallic compound formation or (b) Fe–Al particle formation by cluster intermixing of aluminum and iron elements occurs under growth conditions. The latter are typically nonequilibrium structures, and thermal treatment and surrounding conditions may strongly influence their catalytic behavior.

To confirm the shape and particle morphology of the Fe–Al nanocatalyst, firstly high-angle annular dark-field STEM images were recorded (Figure 1 A). The particles exhibit a core–shell contrast with facets. Their size ranges from a few nanometers up to 50 nm and the particles show sharp corners and edges with flat surface structures. Lattice fringes are observed, offering proof of the crystalline nature of the catalyst particles. High-resolution bright-field tomography of these particles was undertaken to clarify their three-dimensional shape. The tomogram data of the catalyst particles reveal a dome-shaped morphology with a void within the core, hence the donut-like contrast in the projection images. A tomogram slice showing the central void and well-defined facets is shown in Figure 1 B. Spatially resolved electron energy-loss spectroscopy (EELS) was performed in a high-resolution STEM to study the element distribution within the catalyst particles (Figure 1 C). Line profiles of 20 EEL spectra were taken across a number of catalyst particles. The evaluation of core-loss signals in the spectra allows plotting of the intensity of Fe and Al concentrations across the particle’s diameter. As can be seen from Figure 1 C, there is an abundance of Fe and Al in the shell of the particles. Both the Fe and Al contents drop towards the particle’s centre, giving rise to the characteristic contrast in Figures 1 A and B. In summary, the Fe–Al catalyst particle composition is highly anisotropic, but shows no depletion of iron from the particle surface. This result is notable, since it is known that following complete depletion of Fe from the surface, no CNT growth is possible.14

thumbnail image

Figure 1. Microstructure of the Fe–Al catalyst particles: A) High-angle annular dark-field STEM image. Viewed from the top, the particles exhibit a core–shell-like contrast. Fe is accumulated in the shell, reasoning from the approximate proportionality between the high-angle annular dark-field signal and the square of the atomic number; B) cross-sectional view of a single catalyst particle. The image is a slice through a tomogram reconstructed from a tilt series of high-resolution TEM bright-field images. The dotted lines indicate the position of the carbon support of the TEM grid. The catalyst particles have a dome-like shape and their surface is faceted. The bright-field contrast has been inverted for a more intuitive interpretation; C) elemental profiles of Al and Fe, obtained by integration of the EELS signal of the Al K core-loss and the Fe L23 core-loss edges, respectively. The inset in the upper right shows the particle and the profile line, along which 20 locations were probed. The inner part of the particle is a void, deficient of Al and Fe. Error bars indicate the 1 σ uncertainty related to background extrapolation error and signal noise.

Download figure to PowerPoint

This study focuses on the surface chemistry of the Fe–Al catalyst. Al particles (10 nm thickness) were deposited together with Fe (1 nm thickness) on a Si (100) substrate. Three different X-ray photoelectron spectroscopy (XPS) experiments were performed, one heated to CVD temperature of 750 °C without any reactant gases (referred to as sample 1), a second (referred to as sample 2) at the same reaction temperature in an atmosphere of hydrogen, argon, and water, and a third experiment (referred to as sample 3) under similar conditions to sample 2, but with addition of ethylene as a carbon growth precursor (“super long” growth conditions). After CNT growth, the Fe–Al composite catalyst was recovered by peeling off the grown CNT film, allowing a direct comparison of the chemical composition of the catalyst particles in all three experiments, before and after CNT growth.

The chemical composition of the Fe–Al catalyst was probed ex situ for samples 13 with XPS for the elements iron, aluminum, carbon, and oxygen and their core levels provide decisive clues to the surface chemical composition of the bimetallic Fe–Al catalyst particles. For iron, an XPS peak (Fe 2p3/2 at 710.3 eV,) corresponding to iron oxide hydroxide, FeOOH15, 16 is observed (Figure 2 A). Metallic iron if present can probably not be detected by XPS, if it is buried underneath an Al surface layer (Figure 3, discussion on GIXRD). The Fe 2p3/2 peak position did not alter throughout the complete set of experiments (samples 13). Aluminum was found to exist solely in its metallic state in samples 2 and 3 (signal for Al 2p at 72.3 eV, Figure 2 B). Significant oxidic contributions were found only under the nonreducing conditions of sample 1 (signal for Al 2p at 74.9 eV, Figure 2 B), suggesting that for sample 1, partial reaction of the Fe–Al catalyst particles with the native SiO2 layer of the substrate had occurred. Trace surface oxidation due to anaerobic sample transport into the instrument is another realistic scenario, albeit obviously not important for the other samples, since no comparable signals were found there. Figure 2 C shows the evolution of C 1 s XPS peaks for samples 13 under different reaction conditions. The C1 s peak shifts from a minor sp2 C impurity signal at 284.8 eV15 to a major intensity signal at 282.6 eV (sample 2). A further peak shift to 283.4 eV occurred for sample 3, when the experimental CVD conditions had been adjusted to “super-long” CNT growth conditions (Figure 2 C). The C 1s core signal is in accord with an assignment to an aluminum oxycarbide species.17

thumbnail image

Figure 2. X-ray photoelectron spectra showing A) iron, B) aluminum, C) carbon, and D) oxygen core-level lines for the Fe–Al catalyst under different experimental conditions. Corresponding samples are referred to as 1, 2, and 3. Sample 1: heated to CVD temperature of 750 °C; sample 2: heated to CVD temperature in the presence of H2, H2O, and argon; sample 3: heated to CVD temperature in presence of H2, H2O, and carbon precursor ethane. Carrier gas is always argon.

Download figure to PowerPoint

thumbnail image

Figure 3. GIXRD spectra of the Fe–Al composite catalyst particles deposited on a Si substrate, A) before and B) after growth of “super long” CNTs.

Download figure to PowerPoint

The assignment of the corresponding XPS signal to such an Al[BOND]O[BOND]C-bonded species is further substantiated by the O 1 s XPS core signal for sample 3 (530 eV, Figure 2 D), as well as by the grazing incidence X-ray diffraction (GIXRD) studies on the Fe–Al composite catalyst particles after CVD growth (see Figure 3 and discussion).9, 16, 18 In addition, the aluminum oxycarbides, Al2OC and Al4O4C, are well known substances from studies of solidified melts.19

GIXRD gave peaks for metallic iron and aluminum either under conditions when water, argon, and hydrogen but no carbon precursor had been added (Figure 3 A; sample 2). These results, although in contrast to those from the XPS studies, are understandable, since XPS is a surface probe technique and GIXRD allows characterization of the bulk morphology of the Fe–Al catalyst particles. However, when typical reaction conditions for “super-long” CNT growth were adjusted (Figure 3 B; sample 3), characteristic diffraction peaks for aluminum hydroxide and iron oxide hydroxide20, 21 were detected in the bimetallic catalyst phase after growth. Those for metallic aluminum and iron, as present before for sample 2 (Figure 3 A), have disappeared. Thus, when carbon species were present under active growth conditions, the catalyst particles mostly converted to iron and aluminum hydroxide and oxide species, which are detectable in the GIXRD experiment. No diffraction peaks of the intermetallic phase Fe3Al were observed at all.22

Oxygen growth enhancers are crucial in the growth process of “super-long” CNTs.1, 4 They clean the catalyst’s surface of amorphous carbon according to ,

  • thumbnail image

thus allowing steady growth conditions for relatively long periods of time. Therefore water must be adsorbed onto the active catalyst surface during the growth. Size-defined aluminum metal clusters are known to adsorb water as hydroxy groups, thus forming a variety of structurally different aluminum hydroxide clusters.23 Surface hydroxy groups bonded to aluminum are crucial in the growth of “super-long” CNTs. They react with adsorbed carbon species that are formed according to the precursor deposition reaction: C2H4+2 H2[RIGHTWARDS ARROW]2 Csolid+4 H2, giving CO and CO2 (Scheme 1). This process continuously cleans the catalyst’s surface, as long as a sufficient amount of water is supplied.4, 13 In the future, detection of such species in a low coverage regime (few layers to monolayer coverage) on the active catalyst particles for CNT growth would certainly need sophisticated in situ techniques.7

thumbnail image

Scheme 1. Proposed reaction scheme for the activation of the bimetallic Fe–Al catalyst particles in “super-long” CNT growth (O red; H white, C blue). A) The as-deposited Fe–Al catalyst particles react with hydrogen gas and water to form hydroxo species on the active bimetallic catalyst surface at CVD reaction temperature of 750 °C; B) carbonaceous species, formed via decomposition of the molecular precursor ethylene, then react with hydroxy groups at the catalyst’s surface to expel CO, CO2, and H2O; C) in this way the Fe–Al catalyst is kept active during CVD growth conditions, maintaining “super long” CNT growth conditions.

Download figure to PowerPoint

CO and CO2 are formed in an exothermic reaction,24 which help to maintain a strongly negative enthalpy for the “super-long” CNT growth process; ΔH for the above reaction is just −52.5 kJ mol−1, compared to ΔH=−394 kJ mol−1 for C+O2[RIGHTWARDS ARROW]CO2.25

In conclusion, we have shown that the formation of an Fe–Al composite catalyst containing metallic iron and aluminum, as well as mixed oxidic species, are crucial in the water-assisted growth of “super-long” CNTs. The catalyst particles are highly crystalline in nature and they display an anisotropic distribution of both elements across the catalyst particle’s diameter, as evidenced by HRSTEM, EELS, and GIXRD. Based on these analytical findings, a mechanism for ongoing catalyst activation in the water-assisted growth of “super-long” CNTs could be proposed. Our studies presented herein give new insight and understanding of the nature of the chemical composition of the bimetallic Fe–Al catalyst responsible for “super-long” CNT growth. Nevertheless, the nature (active physical and chemical state) of the Fe–Al catalyst is far from being resolved and certainly needs more detailed studies until a more complete picture arises.

Experimental Section

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

CNTs were grown on standard p-type <100> lightly doped silicon with 600 nm silicon dioxide layer. A 10–12 nm-thick buffer layer of aluminum was deposited using electron beam evaporation followed by 1 nm thick iron layer.12 In a similar manner, on carbon-coated copper TEM grids (Fa. PLANO, Wetzlar), a 10 nm thick layer of aluminum was deposited followed by a 1 nm thick layer of iron. For further instrumental and technical analytical details of catalyst and materials characterization see supplementary material.

Acknowledgements

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

Support through the collaborative project ERC-TUD1 is acknowledged with gratitude. A.I. thanks Prof. W. Jaegermann, TU Darmstadt, for the use of the XPS equipment.

Supporting Information

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors.

FilenameFormatSizeDescription
cctc_201000037_sm_miscellaneous_information.pdf2398Kmiscellaneous_information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.