Controlling Energy-Level Alignments at Carbon Nanotube/Au Contacts

The properties of the contacts between single-walled carbon nanotubes (SWNTs) and Au electrodes are studied using scanning Kelvin probe and electrostatic force microscopies. Contact potential differences and local dipoles at the SWNT/Au interface are determined under various conditions involving gas adsorption and surface passivation. In particular, the effects of the coadsorption of alkanethiol, S, and O2 are explored in detail. We find that the coadsorbates alter the energy-level line-up at the contacts and induce significant shifts of the SWNT bands relative to the metal Fermi level. This behavior is explained by considering the response of the local Au work function to the presence of the nanotube and of the coadsorbates as well as the effects of the adsorbate dipoles near the contacts. Finally, we use coadsorption to control the Schottky barrier height at the nanotube−Au contacts. Single-walled carbon nanotubes (SWNTs) are prototypical 1D quantum wires whose electronic properties depend strongly on their diameter and chirality. Thus, SWNTs can be metallic or semiconducting with a band gap that is inversely proportional to their diameter. Scattering and therefore power dissipation are drastically reduced, and their strong covalent bonding allows them to withstand extremely high current densities.1-3 A number of SWNT-based electronic devices have been demonstratedsmost importantly, high-performance SWNT field-effect transistors (SWNTFETs).4-6 In the first studies of SWNT-FETs,4,5 it was observed that they operate as p-type (i.e., hole transport) devices in air, even though they were not intentionally doped. It was later found, however, that in ultrahigh vacuum (UHV) they behaved as n-type (i.e., electron transport) devices.7,8 Initially, the p character of SWNT-FETs in air was ascribed to hole doping by charge transfer from the SWNT to atmospheric O2. However, if the p character of SWNT-FETs in air is due to oxygen doping, then in vacuum the SWNTs should be undoped, and therefore the SWNT-FETs should be ambipolar, not n type as is observed. Recently, however, strong evidence has been presented that the metal-nanotube Schottky barriers determine the electrical character of the SWNT-FETs,7,11-13 and that the transistor switching mechanism in SWNT devices is fundamentally different from that of conventional Si FETs. Specifically, it was concluded that SWNT-FETs are 1D Schottky barrier devices in which the modulation of the carrier injection barrier, instead of the channel conductance, is the basis of the transistor action.11-15 In this model, oxygen adsorption on the metal electrode/ SWNT interface affects the alignment of the nanotube bands with respect to the metal electrode Fermi level, thus altering the carrier injection barriers. As a result, holes are easily injected in air, and electron transport is favored in vacuum. It is clear then that optimizing the performance of SWNTFETs requires a better understanding of the interactions occurring at the SWNT/metal contacts. Unlike the interfaces between metals and bulk inorganic semiconductors, such junctions are poorly understood. Here we report studies of SWNT/Au interfaces with scanning Kelvin probe microscopy16 (SKPM) and electrostatic force microscopy (EFM).17 The results indicate that the interfacial dipole layer at SWNT/Au junctions reversibly changes direction as the environment changes from ambient air to vacuum or an oxygen-free environment. Consequently, the Au Fermi level lies above the SWNT mid-gap in oxygenfree environments and below the mid-gap in air. In contrast, there is no significant dependence of the energy-level alignment at SWNT/graphite (HOPG) interfaces on oxygen. Furthermore, on Au, the energy-level alignment is strongly sensitive to the coadsorption of other molecular absorbates such as H2S or alkanethiol self-assembled monolayers (SAMs). In this way, we obtain a passivation of the contacts * Corresponding author. E-mail: [email protected]. † Columbia University. ‡ Carbon Nanotechnologies Inc. § IBM Research Division. NANO LETTERS 2003 Vol. 3, No. 6 783-787 10.1021/nl034193a CCC: $25.00 © 2003 American Chemical Society Published on Web 04/29/2003 against further oxygen adsorption and a permanent modification of the carrier injection barriers, which provides a way of tuning the transport properties of SWNT-FETs. Examples in which we enhance the performance of p-type SWNT-FETs or convert a unipolar p-type SWNT-FET to an ambipolar one are presented. These results provide further proof of the Schottky barrier model for the SWNT-FETs. Experimental Section. The semiconducting SWNTs used in our experiments were provided by Rice University. They have an average diameter of 1.4 nm18 and a band gap of about 700 meV.19,20 Samples for SKPM and EFM experiments were obtained by dispersing SWNT solutions in dichloroethane onto three different substrates: polycrystalline Au on SiO2, flame-annealed Au(111) on mica (Molecular Imaging), and freshly cleaved highly oriented pyrolytic graphite (HOPG). The SKPM experiments were carried out on a JEOL A4500 microscope using Pt, W2C-coated, or n+Si conducting AFM tips in a UHV chamber (∼10-7Pa) or in air. In the Kelvin probe images (see Figure 1B and C), a bright region indicates a locally lower vacuum level or surface potential. EFM measurements were performed using a Nanoscope III DI microscope with Pt-coated tips (Nanosensor) either in a N2-filled glovebox (O2 ≈ 1 ppm) or in air. In EFM images (Figure 1E and F), the brighter features indicate positive charges or dipoles pointing toward the bulk of the sample (V). In SKPM and EFM measurements, individual SWNT tubes or bundles of tubes were first identified by AFM topographic images. The charge distribution and charge density on SWNTs were deduced by fitting the curves of the first harmonic component of EFM signals, the force gradient, as a function of the tip-sample distance.21 SKPM measures the difference between the local work function of the sample and that of the reference SKPM tip, which is made of n+ Si with nanometer resolution. The EFM measurement, however, provides a measure of the charge transfer that takes place at the nanotube-metal interface. Thus, the SKPM and EFM measurements provide unique information about the electronic structure and interactions