Electronic Transport in Multiwalled Carbon Nanotubes Contacted with Patterned Electrodes Rei HOBARA, Shinya YOSHIMOTO, Takashi IKUNO 1 , Mitsuhiro KATAYAMA 1 , Norihiro YAMAUCHI 1 , Winadda WONGWIRIYAPAN 1 , Shin-ichi HONDA 1 , Iwao MATSUDA, Shuji HASEGAWA and Kenjiro OURA 1 Department of Physics, School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan 1 Department of Electronic Engineering, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan (Received May 24, 2004; accepted June 19, 2004; published July 23, 2004) The electrical conductance of 0:8  5-mm-long multiwalled carbon nanotubes (MWCNT) was measured at room temperature in a multiprobe scanning tunneling microscope (STM)-scanning electron microscope (SEM) system and a conventional prober system, by bringing the MWCNTs into contact with patterned metal electrodes. The contact resistance between the CNTs and metal electrodes was sufficiently small. The conductance was proportional to A=L (and also to B=L, within our experimental error), where A, B, and L are the cross section, circumference, and length of CNTs. This indicates the occurrence of diffusive transport. A nonlinear current-voltage characteristic was obtained; the conductance increased steeply with current. A multiprobe STM-SEM system was very useful for measuring individual CNTs. [DOI: 10.1143/JJAP.43.L1081] KEYWORDS: carbon nanotube, multiprobe STM, electrode, conductance, I -V characteristic The electronic transport properties of cabon nanotubes (CNTs) have received much attention from the points of view of one-dimensional transport physics as well as device applications. 1) Though a number of test devices such as field- effect transistors, 2–4) single electron transistors, 5,6) and nano- sensors 7) have been created with CNTs, the fundamental transport property is still controversial, 8) probably because of their poorly defined quality and structure. Some groups have reported that the multiwalled carbon nanotubes (MWCNTs) exhibit ballistic conduction, 9–11) while other groups have reported diffusive conduction characteristics. 12–14) Another interesting issue is the current path; Collins et al. suggested that the current flows preferentially through the outermost shell of MWCNTs, rather than through the entire cross section. 15) The third point to be addressed is that of the nonlinear current-voltage characteristics; since they have been reported only very recently, their origin has not yet been identified. 16,17) In this Letter, we show systematic measurements of the conductance of MWCNTs prepared by bringing them into contact with metal electrodes at both ends of the MWCNTs in two different ways. One is that the CNTs were dispersed on patterned Ta electrodes, and the conductance of CNTs bridging the two electrodes was measured with a four-tip- scanning tunneling microscope (STM) combined with a scanning electron microscope (SEM) in ultrahigh vacu- um. 18,19) With this system, we could identify the CNTs bridging the electrodes and measure the conductance selectively by touching two STM tips to the electrodes. This method was very effective for identifying and measur- ing many CNTs on a substrate at one time. Another method involved depositing Ti electrodes over the CNTs at both ends. While this technique is expected to be more effective for future device application of CNTs, the contact resistance is said to be problematic in some cases. 9) We found that the conductance g is proportional to A=L, where A is the cross section and L is the length of CNTs, for both methods. The data also appeared to be proportional to B=L within our experimental error, where B is the circum- ference of CNTs. In any case, the results indicates the occurrence of diffusive transport through the CNTs longer than 0.8 mm. But we could not resolve whether the conduction is through the outermost shell (g / B=L) or though the entire cross section (g / A=L), because of the limited CNT diameter range we investigated. We also found nonlinear current-voltage (I -V ) characteristics which were similar to recently reported ones. 16) This phenomenon is, therefore, a property which is intrinsic to MWCNTs, irrespective of the synthesis and measurement methods used. Since our samples exhibited low contact resistance and consistent results, we can say that the contacts at the electrodes were highly reproducible. The experiments were performed in two ways. In the first method (Method I), we used a SiO 2 /Si wafer as the substrate, with patterned square Ta pads separated by various distances (0.8–5 mm). 20) Commercial CNTs having various diameters (0.05–0.2 mm), fabricated by the arc discharge method, were randomly dispersed on the substrate using ethanol or dimetylformamide. We investigated the structure and quality of the CNTs separately by means of transmission electron microscopy (TEM), and confirmed they were of good quality with few defects (Fig. 1(b)). A sample was placed in the four-tip STM-SEM vacuum chamber. 18,19) We chose CNTs which bridged two pads under the SEM, and measured their I -V curves by touching two STM W-tips onto the two pads (Fig. 1(a)). In the case with no CNT between the pads, the resistance was larger than 10 M, and thus was much larger than that between two pads bridged by a CNT. We also confirmed, by touching the two tips onto the same pad, that the resistance between tip and pad was sufficiently small (1). A typical I -V curve is shown in Fig. 1(c). Though it shows slight non- linearity, we took the slope near the origin as the resistance. The energy of our SEM electron beam was 10 keV, and we exposed the beam on CNTs or STM tips for from about 10 minutes to 12 hours, resulting in an exposure of about 20 C/cm 2 at most under usual conditions. The SEM observations showed that the irradiation did not make any visible contaminations or defects on CNT or STM tips (even by 100 times larger exposures than usual), probably because of the ultrahigh vacuum environment in our chamber and room temperature condition. The measured I -V curves and resistance looked unaffected by such electron beam irradi- ations. Some groups have reported 21,22) that CNTs are Japanese Journal of Applied Physics Vol. 43, No. 8B, 2004, pp. L 1081–L 1084 #2004 The Japan Society of Applied Physics L 1081