Reliability and current carrying capacity of carbon nanotubes B. Q. Wei, a) R. Vajtai, and P. M. Ajayan Department of Materials Science and Engineering, Rensselaer Polytechnic Institute, Troy, New York12180 ~Received 16 April 2001; accepted for publication 3 July 2001! The current-carrying capacity and reliability studies of multiwalled carbon nanotubes under high current densities ( .10 9 A/cm 2 ) show that no observable failure in the nanotube structure and no measurable change in the resistance are detected at temperatures up to 250 °C and for time scales up to 2 weeks. Our results suggest that nanotubes are potential candidates as interconnects in future large-scale integrated nanoelectronic devices. © 2001 American Institute of Physics. @DOI: 10.1063/1.1396632# Over the last decade we have witnessed tremendous ad- vances in the characterization of structure and electrical properties of carbon nanotubes ~CNTs!. 1 Interest in control- ling the structure and properties of nanotubes, driven by their unique structures which is directly related to their remark- able physical properties, is accelerating at a rapid pace. The electrical properties of CNTs are fascinating because they can exhibit metallic or semiconducting behavior depending on their structure and dimensions. 2–4 This has made carbon nanotubes a unique candidate material for potential nano- technology applications as nanoscale electronic devices and quantum wires. 5–10 However, there are very few studies yet regarding the reliability of carbon nanotubes in microelec- tronic circuits. Electromigration, referring to current-induced displace- ment of atoms that occurs in a conductive material, is a criti- cal property for interconnect materials ~such as Al, Cu, W, Ag and silicides! in microelectronic circuits. 11 Atoms are dis- placed as a result of combined effects of direct momentum exchange from the moving electrons and the influence of the applied electric field. The current-induced mass transport combined with any divergence in mass flux may cause par- tial removal of material from one location to buildup of ma- terial in other locations, thus resulting in an open circuit in the area in which material is removed ~void formation! or a short circuit in the area in which material builds up ~forma- tion of hillocks or whiskers!. The smaller the material dimen- sion the lower the electromigration resistance for all the ex- isting metals. Thus, conventional metallization lines have limitations in future nanoelectronic ultralarge scale integra- tion applications, especially above certain critical current densities. Multiwalled carbon nanotubes ~MWNTs! can pass a very high current density from 10 6 to 2.4310 8 A/cm 2 without ad- verse effects. 12–17 Our previous results had shown that elec- tron flux of ;10 9 A/cm 2 could be passed through an indi- vidual MWNT at 300 °C for a short time without destroying the nanotube. 18 From the point of view of applications the question is whether they can withstand such high current density for a longer period of time. In addition, theoretical calculations have revealed that only the outer tube in a MWNT contributes to conductance, 19 and this has also been implied by recent experiments. 7,20 Sequential burning of in- dividual nanotube layers at room temperature has been dem- onstrated very recently. 21 Therefore, it is important to look at the long-term stability ~reliability! of nanotubes as parts of electronic circuits, if they are to be used in future micro/ nanoelectronic applications. In this letter we report initial experiments on the long-term behavior and subsequent elec- tromigration resistance of carbon nanotubes. The multiwalled nanotubes used in our experiments were prepared by the electric arc-discharge method. 1 The nanotube powder was dispersed in chloroform using ultra- sonic agitation for 15 min. A drop of this solution was then deposited on an oxidized Si wafer ~with a 200 nm thickness of SiO 2 !. The tungsten leads connecting the nanotubes were prepared by focused-ion-beam ~FIB! lithography ~for details, see Ref. 18!. The prepared samples were mounted onto an operating platform that could be heated. Both the two- terminal and four-terminal resistances were measured on dif- ferent nanotubes in air with the setup, which is computer interfaced to a dc voltage source supply and a multimeter. The samples were tested in constant voltage mode by keep- ing the input voltage at 25 V in two-terminal measurements, and also in the constant current mode with 10 mA main- tained in four-terminal mode. The samples were measured at ambient temperature of 250 °C in air. A typical atomic force microscope ~AFM! image with two tungsten leads connecting an individual carbon nanotube is shown in Fig. 1. The diameter of the nanotubes discussed here are 8.6 and 15.3 nm ~from the height of the AFM pro- file!, and the spacing between the two electrodes is about 2.5 mm. The tungsten electrodes connecting the nanotube depos- ited by FIB lithography have a preset dimension of 200 nm thickness and 100 nm width. The electrode spacings of the probes of the two-terminal and four-terminal measurements were almost the same. Figure 2 shows the stability of the time-resolved resis- tance of two MWNTs, with both two probe and four probe measurements. The two-terminal circuit shows average resis- tance of about 2.4 kV, while the four-terminal one has an average value of about 1.7 kV. The geometrical and electric properties of the samples are summarized in Table I. Inde- pendent measurements showed that connections made by FIB has a very small contact resistance, always below 100 V, and mostly in the range of a few tens of Vs. The most a! Electronic mail: weib@rpi.edu APPLIED PHYSICS LETTERS VOLUME 79, NUMBER 8 20 AUGUST 2001 1172 0003-6951/2001/79(8)/1172/3/$18.00 © 2001 American Institute of Physics Downloaded 14 Aug 2001 to 128.113.37.241. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp