The interaction between carbon nanotubes and metal surface or cluster is of crucial interest for creating innovative nanostructures with superior electrical properties to be employed in applications such as field effect transistors, field emission displays or single electron transistors, where metal contacts can be required to integrate the carbon wires within the device. Moreover the feature sizes in Si integrated circuits are continually reducing and the interconnects between each transistor are requested to carry ever larger current densities. Therefore carbon nanotubes have been also proposed to be used as interconnects at the same time as carbon nanotubes can carry very high current density, before failing as a result of electromigration. Contacts with vanishing Schottky barrier are requested because the overall resistance should be lowered.
It has been shown that gold does not side-wall bind to single wall carbon nanotubes (SWNTs), differently from other metals (palladium and platinum), and this take into account of the large Schottky barrier found in experiments at the Au/SWNT interface. I made first principle calculations in order to explain why palladium does bind with SWNT better than gold.
I started from the principle that the factors controlling the strength of the adsorbate-metal interaction are two:
- the degree of orbital overlap;
- the degree of filling of the antibonding states.
In order to evaluate both factors I proceeded as follows:
My theoretical investigation was carried out with calculations based on the density functional theory (DFT) within the generalized gradient approximation (GGA) with the Perdew, Burke and Ernzerhof (PBE) correlation functional. The calculations were performed using the DMol3 program (from Accelrys Inc.).
I have modeled an armchair (5,5) SWNT with diameter of 6.9 Å either physisorbed or chemisorbed by side-wall to gold or platinum electrode, respectively. The metal surfaces have been modeled by the repeated slab geometry, which contains three atomic layers with the atomic coordinates of the lowest layer constrained. The comparison with the five layer slabs shows a gain of the total energy of 0.12 eV/atom and a shift of the Fermi level (measured with respect to the vacuum level) of 0.03 eV for both Au and Pd systems. Therefore our model is sufficiently accurate to describe the electron hybridization at the interface.
I investigates the orbital overlap between the metal surface and the SWNT hybrid orbitals by studying the partial contributions of metal and SWNT states of different angular momentum to the density of states (DOS). In this way I described how the bonding takes place. Indeed the angular momentum contributing more to the DOS determines the symmetry of the hybrid orbitals. In order to have overlapping orbitals it is requested that metal and adsorbate hybrid orbitals have the correct symmetry, and then metal and carbon atoms can coordinate to form bonds. In our case the formation of covalent bonds is promoted when d orbitals are contributing more to the DOS near the Fermi energy, so that p electrons can coordinate with metal hybrid orbitals to bind the SWNT.
In the case of the Au/SWNT interaction we have a low density of metal d states at the Fermi energy and so the interaction is weak and the SWNT can be only physisorbed.
Strong interaction has been found for the Pd/SWNT interface where the density of states has large contributions from d orbitals so that stable covalent bonds are formed. Indeed orbitals of d-symmetry type are requested in order to have orbital overlapping with carbon p electrons and bond formation.
Filling of the antibonding states
In order to have a clear visualization of the bonding or antibonding character of the orbitals I made the simulations for single atom adsorbed on graphene surface.
The binding energy of the Au atom is much lower than for Pd atom and I found that the HOMO of the Au/graphene system (figure 2) is of anti-bonding character, while the nearest bonding orbital is HOMO-5 for which the energy level is located 1.5 eV below HOMO. On the other side the HOMO-1 for Pd/graphene system (figure 3) is of bonding character and is just 0.3 eV below the HOMO of anti-bonding character. Therefore the Au-C bond is less stable than the Pd-C bond due to the occupation of anti-bonding orbitals; the same considerations can be extended to the interaction between metal surfaces and SWNTs.
In conclusion the different behaviors of gold and palladium surfaces with respect to the side-wall adsorption of the SWNT can be understood in terms of the PDOS analysis at the interface and the degree of filling of the antibonding states. Both factors play a key role in controlling the strength of the adsorbate-metal interaction. In order to evaluate the quality of the generic metal/carbon nanotube contact also other factors should be taken into account, such as the wetting capability and the chemical reactivity of the metal.
The conclusions of this research can be exploited to understand why adherence of organic films deposited above gold substrates is not as good as adherence of other metals, as palladium or titanium.
The full article “First-Principles investigation of the interaction of gold and palladium with armchair carbon nanotube” can be found on Molecular Simulation 36, 729 (2010).