Adv. Mater. 2010, 22, 2177–2181 2010 WILEY-VCH Verlag G Neurophysiologists have used sharpened metal electrodes to electrically stimulate neuronal activities to investigate the physiological functions of the brain. Moreover, they employed this electrical stimulation to treat diseases such as Parkinson’s disease, dystonia, and chronic pain. As neurons utilize electrical potential difference between their cell membranes to transmit electrical signals, this particular way of communication enables us to record the neuronal activity extracellularly or intracellularly. For the extracellular recording approach, the electrodes are positioned intimately next to neuron cells to record and to stimulate their electrical activity by capacitive coupling. The coupling efficacy of these electrical recordings or interventions depends significantly on the selectivity, sensitivity, charge-transfer characteristics, long-term chemical stability, and interfacial impedance between electrodes and target tissue. The most common approach to further investigate the functional behavior of neurons, is using Si-based multimicroelectrode probes fabricated by the micro-electromechanical system (MEMS) method to replace the conventional electrodes (Ag/AgCl) in the aspect of device-structure improvement and scaling down device sizes. However, Si-based MEMS electrodes are extremely rigid and cannot be deformed inside the organs; therefore, the recorded positions are easily shifted and the target tissues are consequently damaged when the animals are in motion. This will become an obstacle in future long-term implantation and real-time recording applications. An alternative method is the use of flexible electrodes presented by several groups. The authors utilized soft materials, such as poly(dimethylsiloxane), SU-8 epoxy-based negative photoresist, and polyimides, to fabricate microelectrodes that can deform while being attached to the tissues and that can also be fabricated into small-scale devices using MEMS methods. Not only would rigid Si-based MEMS probes damage target tissues, the reduced electrode size also resulted in a significantly increase in impedance that may degrade recording sensitivity and limit the stimulating current deliverable through an electrode. In order to resolve above issues, the impedance of the electrodemust be as low as possible. Carbon nanotubes (CNTs) exhibit intrinsically large surface areas (700–1000m g ), high electrical conductivity, and intriguing physicochemical properties. Most importantly, CNTs are chemically inert and biocompatible. Based on the above, the promising advantages of flexible substrates and CNTs lead the attempt of fabricating CNTs directly on flexible substrates as microelectrodes for neuronal recording. In this work, the feasibilities of growing CNTs on flexible polyimide substrates at low temperatures (400 8C) by catalyst-assisted chemical vapor deposition (CVD) and utilizing the above devices (see the schematic image in Fig. 1a and the photo in Fig. 1b) as electrodes for extracellularly neuronal recording were investigated. The electrical enhancement (by UV-ozone exposure), biocompatibility (by neuron cell cultures), long-term usage and adhesion, and the detection of action-potential signals on crayfish (using flexible UV-ozone-modified CNTelectrodes) were examined. After a series of process optimizations, the 5-nm Ni-catalyst layer and C2H2 (60 sccm)/H2 (10 sccm) process gases at 5 Torr were found to be the optimum CNT growth parameters in this work. Besides, the Au layer could facilitate CNTgrowth. Figure 1c shows that CNTs have been grown on the polyimide substrate with Au layer, while not on that without Au layer (the inset). The high-resolution transmission electron microscopy (HRTEM) image (Fig. 1d) further confirms the successful syntheses of multi-walled carbon nanotubes (MWCNTs) at 400 8C or even down to 350 8C with H2 plasma pretreatment prior to the CVD processing. As shown in the Supporting Information (Fig. S1a),