Bitte wählen Sie ihr Lieferland und ihre Kundengruppe
The formation of anti-phase disorder during MOVPE growth of pseudomorphic GaP films on Si(100) substrates served as a lattice matched model system for the crucial IH-V/Si(100) interface. A variety of surface-sensitive methods was applied to establish suitable Si(100) substrate preparation and subsequent GaP growth free of anti-phase domains (APDs), by analyzing the substrate surface, the interface and the epitaxial film resulting from the heteroepitaxial process. We studied thermal removal of native and protective wet-chemical SiO2 layers from Si(100) substrates in a MOVPE reactor by means of XPS and required annealing at a surface temperature of 950 °C for reliable deoxidation in a hydrogen ambient. This critical surface temperature for oxide removal was significantly higher than that in established UHV procedures and could be decreased by lowering the process pressure or by supplying silane, while a slight increase in argon-based processes was observed and the alternative use of nitrogen as process gas induced silicon nitride formation. Using LEED, STM and AFM, we found distinct two-domain (2 x 1)/(1 x 2) reconstructions with an equal distribution of smooth SA type and frayed SB type monolayer steps on Si(100) samples with 0.1 and 2 off-cut. In contrast, the prevalence of a single domain and a significant fraction of double layer DB steps were detected on 6 surfaces. We were able to obtain Si(100) surface structures with DA-like steps by the growth of a homoepitaxial buffer layer even on 0.1° misoriented samples. However, in most places, high resolution STM images revealed intermediate SB steps only a few dimers long. Dependent on the rate of cooling, we also observed DA-like steps on the 2 Si(100). Although intermediate SB steps were visible, in several places true DA steps seemed to prevail. Since our findings contradict established UHV results for Si(100) surfaces with and without hydrogen coverage, we investigated the interaction of hydrogen from the process ambient with the Si(100) surfaces. ATR mode FTIR spectra showed the characteristic coupled stretch modes of H-Si-Si-H dimers and tipinduced STM desorption experiments confirmed a complete monohydride termination after processing. However, in situ RAS characterization at typical surface preparation temperatures indicated effectively hydrogen free Si(100), since desorption processes probably outbalance hydrogen adsorption. The in situ analysis of heteroepitaxial GaP films on Si(100) requires a detailed understanding of the surface reconstruction mechanisms involved. The atomic structure of P-rich reconstructed GaP(100) typical for MOVPE preparation was characterized by STM. The images resolved rows of alternatingly buckled H-stabilized P-dimers arranged either in phase or out of phase, which refer to (2 x 2) and c(4 x 2) domains of the surface reconstruction, respectively. The investigations revealed a dimer flipping mechanism, which could also be found on P-rich prepared InP(100). Flipping events switch the tilt of individual dimers and may locally leed to a transformation between the two possible P-rich surface reconstructions. In general, GaP(lOO) and InP(100) surfaces behave very similar, in particular during MOVPE preparation using hydrogen as the standard process gas, but we found and characterized an additional surface reconstruction regime for preparation nitrogen preparation of GaP(100), which does not exist for InP(100). To permit quantitative signal analysis, the impact of changes of the sample temperature as well as the reconstruction and the atomic order of the GaP(100) surface on the assigned RAS signatures were considered in separate experiments.