Thiruvelu
Bhuvana‡
a,
Kyle C.
Smith
b,
Timothy S.
Fisher
*b and
Giridhar U.
Kulkarni
*a
aChemistry and Physics of Materials Unit and DST Unit on Nanoscience, Jawaharlal Nehru Centre for Advanced Scientific Research, Jakkur P O, Bangalore, 560 064, India. E-mail: kulkarni@jncasr.ac.in
bDepartment of Mechanical Engineering and Birck Nanotechnology Center, Purdue University, West Lafayette, IN 47907-2057, USA. E-mail: tsfisher@purdue.edu
First published on 16th September 2009
An easy and elegant method of CNT nanocircuit fabrication using a metal organic precursor of Pd, namely, Pd hexadecanethiolate, is presented. This precursor directs the self-assembly of individual CNTs spanning a gap between Au electrodes. This is achieved by first patterning the precursor along the edges of the gap electrodes, as it enables direct patterning by e beam. Further, thermal activation of the precursor at 250 °C leads to metallization and the ohmic electrical contact between the CNTs and the electrodes beneath. A resistive fuse action of the soldered CNTs is observed as well.
Among the investigated methods for contacting CNTs, one may differentiate those in which individual or a set of CNTs are addressed from a random or an orderly collection spread over a substrate, from those wherein CNTs are deposited only onto desired locations. The former methodology is time intensive as it involves multiple steps: locating a CNT using a microscopy tool, marking its location and finally depositing electrodes. This is usually achieved by performing electron (or ion) beam induced metal deposition (EBID or FIBID) facilitated by a scanning electron microscope or by shadow masking prior to physical vapor deposition.3–6 Instead of metal deposition, there are a few reports where movable microcantilevers served as contacting electrodes.7,8 Langford et al.9 have discussed the various methods for electrical contacts on CNTs, and Yaglioglu et al.10 have studied means of characterising sheet and contact resistance.
Several methods have been reported in the literature for the deposition of CNTs at desired locations, using approaches such as chemical functionalisation and self-assembly. For instance, Lewenstein et al.5 fabricated circuits by amino-functionalising the metal electrodes whereas LeMieux et al.11 functionalised the gap between the electrode. Klinke et al.12 used functionalised CNTs which were placed in designated areas, and electrodes were built subsequently over each adsorbed CNT. Using AFM for manipulation, Gao et al.13 placed an individual CNT between two opposing metal electrodes and realised a four-point arrangement using two additional CNTs. Li et al.14 made use of a PDMS microchannel mold and directed SWNTs into channels under gas pressure to transfer them between a pair of gap electrodes. Using a catalyst bed beneath an anodised alumina membrane, Maschmann et al.15 successfully grew vertical SWNTs through the pores and established top contact by electroplating Pd nanocubes. Another commonly employed method is dielectrophoresis, in which the electric dipole of a CNT exposed to an inhomogeneous electric field is made to guide a CNT to a specific electrode location.16,17
Despite these innovations to make CNT circuits, contact resistance reduction at the CNT-electrode interface remains an active area of research. Bachtold et al.18 exposed a prefabricated CNT-Au electrode system to the electron beam in a SEM chamber to improve the contact resistance between a CNT and Au electrode. The contact resistance was reduced by several orders of magnitude due to exposure to the e-beam. EBID19,20 has also been employed to create metal deposits at CNT-electrode junctions effectively to serve as solder material. Though such deposits are not truly metallic,21 they do improve contact. In another study, Au nanoparticlesol was used as an ink to write as a ‘fountain pen’ at contacts.22 Contacts between CNTs and metal electrodes can also be improved by rapid thermal annealing.23
The present work is aimed at building single CNT circuits, primarily through a self-assembly process in contrast to those previously discussed. Electrode regions are selectively coated with a metal precursor, Pd hexadecanethiolate, employing a direct write method,24 which motivated this work. Following the deposition of CNTs, the precursor is easily metallised by thermolysis to serve as a solder in situ, thus establishing a reliable ohmic contact.
![]() | ||
Fig. 1 Scheme of the procedure adopted. (a) Au gap electrodes, (b) Pd hexadecanethiolate is patterned on Au electrodes along the edges, (c) after dropping the CNT dispersion, the CNTs self-assemble across the gap and (d) the CNTs are soldered by simply activating the patterned thiolate. |
![]() | ||
Fig. 2 SEM image of the Au gap electrodes after drop casting with CNTs along with its schematic (a) with and (b) without Pd hexadecanethiolatepatterning along the electrode. In (a), zoom-in images are shown as insets. SEM images were taken in 0.1 Torr of water vapor/N2 (environmental mode) in order to avoid local charging of the glass substrate. While electrode regions are clear, some local charging is seen in the gap region. |
The above observations are indeed striking given that this self-assembly involves no external direction; such a circumstance is highly desired for CNT-based circuits. It is also surprising to see several individual CNTs across the gap electrodes while the dispersion itself was found to contain, as expected, highly entangled bundles of CNTs. In the absence of Pd hexadecanethiolatepatterning, CNTs deposited over Au electrode regions have disordered orientation (see Fig. 2b). The test with the patterned precursor, Pd hexadecanethiolate, clearly involves a self-assembly process. Following repetitive experiments, we have found that nearly 80% of the tubes in the vicinity of electrodes (i.e., within 10 µm) assemble across gap electrodes; otherwise, the tubes exhibit random orientation such as shown on the left side of the SEM image in Fig. 2b. Similar observations have been reported on PDMS stamp surfaces,25 where CNTs self-assemble across stamp microchannels. In this case, a similar mechanism is likely active whereby the channel region forms a pool of solvent that is the last to evaporate. As the liquid front recedes into this channel, alignment of CNTs occurs such that they are perpendicular to the front and thus finally bridge the channel because the CNT length exceeds the size of the channel width. Further, in this case, Pd hexadecanethiolate is itself a hydrophobic self-assembling molecule,26 and consequently, its presence may produce preferential surface interactions with the hydrophobic CNTs. The present approach offers an additional and important advantage that the precursor can be converted to an electrically conductive contact as described in the following.
![]() | ||
Fig. 3 I-V characteristics of (a) CNTs self-assembled across the gap and (b) after thermolysis at 250 °C for 10 minutes. |
![]() | ||
Fig. 4 (a) SEM image of a CNT on the cross-linked Pd precursor (marked Pd) on Au electrodes (the gap is blurred due to local charging while imaging). (b) I-V characteristics after thermolysis at 250 °C for 10 minutes. |
The CNT circuits thus fabricated are indeed nanometric resistors. The Joule heating of CNTs is typically of the order of 10–100 µW and is enough to cause a temperature rise of hundreds of degrees Celsius.28;29 Under such conditions, CNTs can undergo burn out. One such example is the circuit shown in Fig. 2, with examples of CNT-based fuse action in Fig. 5a and b. In these examples, the CNTs seem to burn out in the middle due to high current. The observation of failure in the CNT device region (Fig. 5a and b) rather than in the contacts further indicates the low electrical resistance of the contacts and diffusive nature of the electrical conduction;30,31 in addition, some favorable annealing may occur in the contact regions under self heating such that the interface resistance decreases both electrically and thermally. Under uniform self-heating of a cylinder with negligible electrical contact resistance, the maximum temperature rise will occur at its centre. The fact that the observed failure occurs very near the exact centre of the CNT device suggests that the heat generation profile is relatively symmetric. The magnitude of maximum temperature rise depends on local electrical heat generation, CNT diameter, contact width between the CNT and substrate material in the gap between electrodes, span of the gap, and specific thermal contact conductance in the gap. Using a simple diffuse mismatch model to estimate the specific thermal contact conductance32 and calculating contact width with a plane-strain adhesive contact model,33 we estimate that a local heat generation rate per unit CNT length of 120 W/m is required to bring about air oxidation of a 150 nm diameter CNT centrally lying on a 9 µm SiO2 gap (assuming that the thermal oxidation in air takes place at 700 °C34). Local heat generation depends strongly on applied bias and the temperature-dependent resistivity of a CNT. Future measurements of CNT resistivity will aid evaluation of the model and in quantitative prediction of breakdown voltages.
![]() | ||
Fig. 5 SEM images of blown off CNTs from Fig. 2a while applying excess external bias. |
Multi-walled carbon nanotubes (Sigma Aldrich) of diameters 150–300 nm and lengths 7–12 µm were used as active elements across the gap electrodes. The CNTs were dispersed in carbon tetrachloride solution by ultrasonication for 30 minutes. A 10 µL amount of the CNT dispersion was drop-cast upon the patterned substrate. Carbon tetrachloride was chosen as the medium for nanotube dispersion, not just because the nanotubes form a good dispersion in it but also because the patterns do not get washed away as the Pd hexadecanethiolate is insoluble in this solvent. In a few standardisation experiments (see ESI† ), carbon fibres (diameter 7–8 µm and length ca. 5 mm) were placed between Au electrodes on glass plate (gap width ca. 1.1 mm) using forceps, and Pd hexadecanethiolate was either spin-coated prior to fibre placement or drop-coated on top. In all current–voltage characteristics measurements, a Keithley 236 multimeter served as the source and measurement unit.
Footnotes |
† Electronic supplementary information (ESI) available: Details of the thermolysis of Pd hexadecanethiolate and preliminary results on carbon fibre as active element between the gap electrodes. See DOI: 10.1039/b9nr00035f |
‡ Present address: Purdue University, West Lafayette, IN 47907-2057, USA. |
This journal is © The Royal Society of Chemistry 2009 |