DOI: 10.1002/adma.200701146 Direct Growth of Flexible Carbon Nanotube Electrodes** By Jun Chen, Andrew I. Minett, Yong Liu, Carol Lynam, Peter Sherrell, Caiyun Wang, and Gordon G. Wallace* The abundance of carbon, together with its accessibility, pro- cessability, chemical stability, and wide electrochemical poten- tial window are reasons why carbon, in all its allotropes, repre- sents a very attractive material for use in energy conversion and storage. It has found widespread use as an electrode mate- rial in capacitors or batteries, [1,2] and even as component of fuel cells. [3] Common energy-storage devices such as batteries and capacitors rely on large-surface-area electrodes to function. [4–6] Hence, having a high electroactive surface area combined with high conductivity and useful mechanical properties makes car- bon nanotubes (CNTs) attractive candidates for electrodes in these devices. Following the successful synthesis of aligned CNT forests, [7–10] carbon nanotube electrochemical devices came into sight, [11,12] and several chemical vapor deposition (CVD) methods have since been developed to grow aligned multiwalled and single-walled carbon nanotube forests. [13,14] However, while impressive results were reported, these ap- proaches suffer from the inherent problem that growth must be established on nonconducting substrates. Once synthesized, the aligned forests need to be transferred to a conducting sub- strate, [15] or require a post-processing step where metal con- tacts are deposited on top of the forest, [16] before they are suit- able as electrode materials for integration into devices. More recently, direct growth of CNTs on Ni-containing alloys was re- ported, and the obtained materials were subsequently effec- tively used as capacitors or field-emission devices. [17] Addition- ally, the growth of CNTs on metals using plasma-enhanced CVD was reported, but this technique has still issues of scal- ability. [18,19] CNTs have commonly been grown from metallo- cenes of Fe, Co, Ni, Ru, and/or metal oxides, nitrates, or penta- carbonyls Much less common in the literature is the use of organic iron salts as catalysts, such as iron(III)-tosylate (Fe(III)TS), iron(III)-dodecylbenzenesulfonate (Fe(III)DBS) and iron(III)-pyridinesulfonate (Fe(III)PS). Here, we report that by modifying the generally employed route to CVD synthesis it is possible to grow carbon nanotube networks integrated into a carbon layer (CL) on insulating gra- phitic carbon (glassy carbon) and even on metallic substrates (e.g. aluminum or copper foil). Although networks derived through this new process contain unaligned multiwalled nano- tubes only, they are vastly superior in all electrochemical as- pects when compared to vertically aligned forests grown in the same furnace. This, combined with the robust nature and flex- ibility of the structures produced, provides an exciting advance. In a typical experiment (see Experimental Section), a thin film of iron(III) p-toluenesulfonate (Fe(III)pTS) catalyst was spin-coated onto quartz plates from organic solutions. After annealing, which facilitates solvent removal, the CVD process was initially carried out at 600 °C under Ar/H 2 gas flow to re- duce the Fe 3+ catalyst to iron nanoparticles. A CNT growth phase followed, carried out at 800 °C with C 2 H 2 as the carbon source. The resultant CNT films (Fig. 1a) were observed to be very much unlike those grown using conventional catalysts. During CNT growth a reflective layer was formed beneath the carbon film (Fig. 1b). The formed carbon nanotube/car- bon layer (CNT/CL) paper appears as a matt-black layer on top of the quartz plate (Fig. 1a), while a flexible, shiny carbon layer forms the lower layer of the CNT/CL paper (Fig. 1b). The CNT/CL paper was easily removed from the substrate and the resulting freestanding film could be rolled around a glass rod without visible signs of degradation (Fig. 1c). The size of the CNT/CL 3D networks is easily scaled up to 100 cm 2 by using a larger CVD quartz tube, given that scale up is not limited by the CVD process itself. The CNT/CL paper was characterized using a range of methods. Scanning electron microscopy (SEM) (Fig.1d) and transmission electron microscopy (TEM) confirmed the top layer indeed consisted of CNTs, whilst SEM of the cross-sec- tional area (Fig. 1e) revealed a highly porous 3D structured CNT network grown on top of a dense CL with less than 1 lm thickness. The long nanotubes obtained in the network are multiwalled carbon nanotubes (MWCNTs) with an external diameter of 20–40 nm. An SEM image of the CL (Fig. 1f) dis- plays a uniformly dense continuous film. Raman spectroscopy (Supporting Information, Fig. S1) of the nanotube layer pro- duced D- (1347 cm –1 ) and G-bands (1598 cm –1 ) within the ac- cepted literature range for MWCNT samples. [21] XRD spectra of the carbon layer revealed peaks in the 2h degree regions of 25° and 42°, identical to those obtained for a commercial Car- bon Black sample (Fig. S2). The electrical resistance was mea- sured using a standard four-probe system, with the CNT/CL COMMUNICATION 566 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2008, 20, 566–570 – [*] Prof. G. G. Wallace, Dr. J. Chen, Dr. A. I. Minett, Y. Liu, Dr. C. Lynam, P. Sherrell,Dr. C. Wang ARC Centre of Excellence for Electromaterials Science, Intelligent Polymer Research Institute, University of Wollongong Northfields Avenue, Wollongong, NSW 2522 (Australia) E-mail: gwallace@uow.edu.au [**] We thank the Australian Research Council for continued financial support, and Jon Lai and Lam Vuong from Atomate USA for helpful discussions regarding the Atomate Thermal CVD. Supporting Infor- mation is available online from Wiley InterScience or from the author.