Chapter 7

Conclusion

The clean surfaces of elemental metals and semiconductors embody a rich field of study. These surfaces relax and reconstruct in order to minimize the surface energy, given the new chemical environment of the semi-infinite crystal. On metallic elements, the surface seeks to smooth its electronic charge density, that otherwise would extend into the vacuum, and the surface atoms respond to their reduced coordination of Coulombic potentials. Most clean metal surfaces relax with damped contractions and expansions of the interplanar spacings, but some also reconstruct when the intraplanar forces are stronger than the interplanar forces. Semiconductor surfaces are more complex; the bonding between atoms is directional and covalent. The truncation of a semiconducting infinite-crystal creates dangling bonds. Hence, complex reconstructions and relaxations occur which redistribute the surface atoms in order to minimize the number of dangling bonds. However, the reconstructions are limited by the ability of the atoms to rehybridize, alter bond lengths, and bend bond angles. Hence, the creation of adsorbate superstructures on semiconductor surfaces is also limited, as typified by the surface phases of B, Mg, Au, and Ce on Si{111} summarized next. Four appendices which briefly describe the known or contested structural information for other Si{111}rt3*rt3-30-X phases follow this chapter.

We expect the surface structures of group-III and -IV elements on Si{111} to consist of well-defined covalent bonds and to satisfy the Si dangling-bonds. Al, Ga, In, Sn, and Pb satisfy such expectations by adsorbing on the bulk-terminated surface above every third second-layer Si atom. This site, called T_4, satisfies the dangling bond of three Si atoms, and a resultant Si{111}rt3*rt3-30-X structure occurs. Boron, on the other hand, is a small atom. If it were to occupy a T_4 site, then substantial stresses would be created. To avoid this stress, boron substitutes for every third second-layer Si atom; the replaced Si atom resides in the T_4-site and transfers charge to the underlying boron atom. This arrangement, called B_5, produces a rt3*rt3-30 surface whose stability is demonstrated by a lack of dangling bonds (as opposed to the clean 1*1, 2*1, and 7*7), by a high heat of formation, and by chemical inertness. An elastic energy-analysis confirms the stability of the B_5 structure for small adsorbates and the T_4 structure for large adsorbates. We have determined the structure (i.e., the distances between all atoms in the top two double-layers) of the Si{111}rt3*rt3-30-B phase with an analysis of low-energy electron diffraction-spectra. The structure we have determined agrees well with that determined by other surface structural analyses.

Magnesium presents an altogether different case. Since Mg is a group-II element with a stable silicide of the CaF_2 type, we are suspicious of a reported Si{111}rt3*rt3-30-Mg phase with a 1/3-ml T_4-structure. This is because of the required three Mg-Si bonds per Mg atom. On the other hand, the reported close-packed hexagonal layer, similar to a hcp Mg(0001) basal-plane, forming on Si at 4/3 ml of Mg does seem plausible. The proposed structure for this rt3*rt3-30 phase is different from the two rt3*rt3-30 structures determined for the close-packed layers of Pb on Si{111} and Ge{111}. We have not been able to prepare either Si{111}rt3*rt3-30-Mg phase, but we have succeeded in preparing a Si{111}(2/3)rt3*(2/3)rt3-30-Mg phase, which we associate with a reacted SiMg_2 overlayer. We have also created a Si{111}3*1-Mg phase, which has diffraction spectra remarkably similar to those of Si{111}3*1-Li, -Na, and -Ag. These similarities can be justified by assuming that charge is exchanged between adsorbate and silicon atoms, but the metal atoms are not ordered in the surface unit-cell. This complex phenomenon requires further study.

Gold on Si{111} surfaces typifies a yet completely different case. Gold does not readily form a silicide, is not a dopant, and forms complex interfaces when ohmic contacts are created. Several surface-sensitive techniques have been applied to the study of the Si{111}rt3*rt3-30-Au phase and the most important finding of each technique was confirmed by our LEED intensity analysis. Ion scattering informed us of missing Si atoms, x-ray diffraction informed us of the Au-Au distance and of the restriction of certain registrations, scanning tunneling microscopy informed us of the simple topology, photoemission restricted the surface symmetry, and total-energy calculations provided us with a complete model. With all of the above information, our LEED intensity analysis confirmed a structure which is missing the top layer of Si, is trimer-based, has mirror symmetry, and contains very little substrate relaxation. We hope that further LEED-intensity calculations will examine deeper Si-layers and will allow us to determine the 5*1 and 6*6 structures.

Cerium on Si{111} presents the case of silicide formation, as opposed to adsorbate superstructures followed by metal thin-film growth. Since all the rare-earth elements form silicides (in several polymorphic structures), we are not surprised by our inability to create the reported simple rt3 and 2*2 phases. We believed that these phases were probably misidentified by previous researchers.

We can now see why metal-semiconductor interfaces are very interesting. The transformation of the stable Si{111}7*7 surface, created by high temperature annealing, to the Si{111}rt3*rt3-30-metal surface, created by metal atoms incorporated into the topmost atomic layers, embodies a rich field of study for surface science. Knowledge of the atomic structure and the coordination of atoms in these rt3 systems clearly play a role in understanding the chemical and electronic properties of semiconductor systems.

The list of well-known surface structures of Si{111}rt3*rt3-30-X caused by the adsorption of Al, Ga, In, Sn, Pb, Sb, and Bi (briefly described in the following appendices) has now been lengthened by the Si-rt3-B system and by the Si-rt3-Au system. Further structure analyses are needed to confirm the structures of Si{111}rt3*rt3-30-Ag and -Si, as well as those for Ge{111}rt3*rt3-30-X phases.