Daniel Gall
The Condensed Matter & Surface Sciences Colloquium Series presents Daniel Gall of Rensselaer Polytechnic Institute on “Materials Property Guessing: The Resistivity Problem,” on Thursday, Nov. 5, at 4:10 p.m. in Walter 245.
Abstract: We can guess materials properties like the density or the melting point, but we (or at least I) fail completely in guessing the electrical resistivity of even of the simplest metals. In this talk, I will present the challenge of predicting the resistivity of metals, particularly for the case of narrow (~10 nm) wires where electron scattering at surfaces and grain boundaries dominate. We use vacuum deposition techniques to grow epitaxial Cu, Ag, Ta, Ti, Ni, W, and TiN layers to study electron scattering at surfaces and interfaces using in situ transport measurements. Atomically smooth Cu(001) surfaces yield specular electron scattering in vacuum, but the resistivity increases when adding a monolayer of Ta or exposing the surface to oxygen, indicating that electron surface scattering becomes completely diffuse. Opposite trends are observed for Cu-Ni, Cu-Ti, and Cu-TiO2 interfaces, showing that electron scattering is controlled by the shape of the interfacial potential drop as well as the add-layer density-of-states.
The resistivity of metal layers increases with decreasing thickness. Fitting the measured data provides a characteristic (metal and temperature dependent) length scale which is associated with the bulk electron mean free path, as described by the classical Fuchs-Sondheimer model. This works perfectly for metals with nearly spherical Fermi-surfaces like Cu or Ag, but the measured mean-free-path disagrees completely (wrong by an order of magnitude) with the free electron model for other metals including Ta, W, TiN. First-principles simulations which correctly account for the shape of the Fermi-surface improve over the free-electron prediction and also indicate anisotropic electron scattering effects, but cannot explain the large disparity between experiment and model. In addition, the measured temperature-dependence disagrees with the existing models and first-principles electron scattering calculations indicate that bulk and surface scattering events are coupled, suggesting that a new, yet to be developed model is required to predict the resistivity of 10-nm-wide wires.
Comments