Reports: AC5

46323-AC5 Identifying the Driving Forces for Alloying in Ultra-Thin Films

Karsten Pohl, University of New Hampshire

The goal of our project is to determine surface structure and composition, and to explore kinetic and dynamical surface processes at a nanometer scale. The approach used is our newly developed LEEM-IV (low-energy electron microscopy – intensity vs. voltage) technique, which is based on conventional LEED-IV (low-energy electron diffraction – intensity vs. voltage) performed on a LEEM system and delivers an in-plane resolution of about 8 nm techniques. This new method allows therefore for the first time the truly three-dimensional compositional analysis of heterogeneous nanostructured surfaces. Several projects have been performed and accomplishments have been achieved as described below. We have made significant progress toward the ability of engineering high-performance surface alloys for catalytic and electro-magnetic applications.

Pd-Cu(100) surface alloys are interesting as model systems for metal/metal epitaxy, as well as for their catalytic properties, and as coatings, e.g. for electromigration resistance. We employ the LEEM-IV technique, with 8.5 nm spatial resolution and submonolayer chemical sensitivity, to investigate Pd interdiffusion into the Cu(100) surface. The LEEM-IV technique is sensitive to the layer-by-layer composition down to the fourth subsurface layer. After annealing a 0.4 ML Pd surface alloy at around 540 K, some regions of the surface develop a Cu3Pd structure, a familiar bulk alloy phase. In other regions, the surface Pd concentration becomes dilute due to Pd diffusion into the bulk. We estimate the thermal activation barrier to Pd diffusion from the surface alloy into Cu bulk to be 1.7±0.15 eV. The LEEM allows real-time, real-space, observation of the interdiffusion process, and the concurrent evolution of the surface structure, at the nanometer scale.

SiC surface is a good platform to make an atom layer of graphite – graphene. In our study, the surface phase transition of Si-terminated 6H-SiC(0001) upon heat treatment is studied by low energy electron microscopy (LEEM). Bright and dark field imaging demonstrates a direct in situ observation of the surface phase evolution, transitions in a sequence from 1×1, 3×3, Ã3×Ã3, 6Ã3×6Ã3 to the graphene phase due to gradually increasing the temperature. Intensity vs. voltage (IV) spectra extracted from single domain diffraction images is used to determine the local surface structure and chemical stoichiometry. Results from a quantitative dynamical analysis of the LEEM-IV curves show a Si-depleted 1×1 structure and an adatom-trimer-adlayer structure on 3×3 reconstruction. We have learned how to make cubic structure on the Ã3×Ã3 structure in this investigation. Ongoing work on the structure of the Ã3×Ã3 and 6Ã3 × 6Ã3 phases is aimed to unraveling the initial growth mechanism of graphene on SiC.

The Bi(001) surface is unique in its surface electronic property. While all Bi surface states studied are spatially confined to the first layer, Bi(001) is a notable exception with deeply penetrating states, which could have a significant effect on the bulk properties of nanostructures. This work concerns surface morphology observation by STM and atomic structure determination by LEED, which are expected to be closely related to the electronic properties. STM shows an unreconstructed surface and wide terraces with double-layer step heights of about 3.76 ± 0.02 . We also identify the short termination by obtaining unstable single step heights via special sputtering operations. In the LEED analysis, the termination with an intact bilayer also results in a much better agreement between calculated and measured intensities than the broken bilayer. Strong multilayer oscillatory relaxations (about 10%) are found to reach deep into the fifth layer, which can be seen as the structural response to the unusually deep surface state penetration at this surface. The measured relaxations agree well with those from first-principles calculations.

Currently we are working on overcoming a serious theoretical challenge in applying this powerful new technique of LEEM-IV to graphene and other directionally bonded materials, such as semiconductors.  The dynamical analysis of the electron reflectivity that delivers the most valuable surface information in LEEM-IV is traditionally carried out using the multiple scattering theory introduced by Pendry.  The multiple scattering approach in principle is not limited to crystal potentials with local spherical symmetry, but muffin-tin (MT) potentials have been widely used for surface crystallography due to computational efficiency.  However, the corrections due to the actual crystal potential in the interstitial region outside the MT sphere can be significant particularly for very low energy electrons.  Indeed, we have observed less than perfect agreement between the calculated and the measured IV spectra for epitaxial graphene layers grown on SiC(0001) for electron energies below 50 eV.  We are proposing to improve the reliability of our LEEM-IV analysis technique by implement a Bloch wave matching method to interpret diffraction curves.  The idea of matching Bloch waves can be traced back to the earliest theories of electron diffraction by Bethe and Laue an it was first applied to the LEED-IV structural determinations by Marcus and Jepsen.

We are using the density-functional approach to determine the Bloch waves which takes into account the full crystal potential directly.  A slab geometry is used for layered structures such as graphene.  The electron scattering amplitudes are determined by matching the Bloch waves to plane waves at the boundaries.  The vacuum region between slabs will be taken to be large enough to eliminate the effects from neighboring slabs.  Similar approaches to interpreting electron diffraction spectra with band structures have been studied, but they were based on matching plane waves with the Bloch wave functions in the bulk.  In contrast, our method focuses on the matching in the vacuum region between well separated structures, which not only avoids the use of complex wave vectors, but also provides a more realistic description of the wave functions near the surface.  So far we have tested simple scattering potentials, such as a periodic 1D quantum wells, and found that our wave matching method reproduces the analytical transmission and reflection coefficients. It is worthwhile to mention that the rapidly increasing computing power is important to the feasibility of our method.