AVS 49th International Symposium
    Surface Science Thursday Sessions
       Session SS+EL-ThM

Paper SS+EL-ThM2
A LEEM Study of the Ge(001)-(2x1)-(1x1) Phase Transition; Domain Wall Proliferation and Dimer Break-up

Thursday, November 7, 2002, 8:40 am, Room C-110

Session: Structure of Semiconductor Surfaces & Interfaces
Presenter: B. Poelsema, University of Twente, The Netherlands
Authors: E. van Vroonhoven, University of Twente, The Netherlands
H.J.W. Zandvliet, University of Twente, The Netherlands
B. Poelsema, University of Twente, The Netherlands
Correspondent: Click to Email
The Ge(001) surface exhibits two phase transitions. At low temperatures it is c(4x2) reconstructed, evolving with increasing temperature into (2x1). The origin of the reconstruction is dimerization of the surface: the number of dangling bonds is reduced from two per surface atom, for a bulk terminated surface, to only one. In the c(4x2) phase the dimers are buckled in an anti-symmetric way; in the (2x1) phase the dimers rapidly switch between the two buckled orientations and appear symmetric. Due to the diamond structure of Ge, the dimer rows on neighboring terraces are rotated by 90°. At high temperature the (2x1) phase disappears and the (1x1) phase emerges. Two conflicting models have been proposed in literature. One model suggests that this phase transition is driven by vacancy pair creation and dimer break-up on the Ge(001) surface. The other claims that the phase transition involves (2x1) domain wall (step) proliferation rather than dimer break-up. Our results demonstrate that domain wall proliferation sets in around 950 K, leading to a complete loss of contrast in LEEM around 1050 K. The dimers, however, remain clearly visible up to about 1130 K. The dimer concentration is a strong function of the substrate temperature between 1030 and 1130 K. Our combined microscopy and diffraction data are only consistent with the first model. It is even possible for the first time to directly extract the free energy gain of dimerization, being 1.6 eV per pair. This value compares perfectly with calculations performed for silicon after scaling with the melting temperature. We estimate the temperature to be accurate with ± 25 K and thus the dimerization energy with ± 5%.