Revisiting the vibrational spectra of silicon hydrides on Si(100)-(2x1) surface: What is on the surface when disilane dissociates?

Even though the decomposition of disilane on silicon surfaces has been extensively studied, the molecular mechanism for its decomposition has not been fully resolved. The general view motivated partly by spectroscopic data is that decomposition occurs through silicon-silicon bond dissociation although there is evidence from kinetics that silicon-hydrogen bond dissociation is important, and perhaps even dominant. Thus, we reexamine the assignment of the experimental vibrational peaks observed in disilane and silane adsorption in order to assess the evidence for the silicon hydride species that are formed during decomposition. We calculate the vibrational density of states for a number of silicon hydride species on the Si(100)-(2x1) surface using Car-Parrinello molecular dynamics. We obtain the calculated vibrational frequency in the adiabatic limit by extrapolating to zero orbital mass, calibrating our method using the well-established monohydride peak. The calculated vibrational frequencies of the monohydride are in good agreement experimental data. Our results show that the spectroscopic data for silicon hydrides does not preclude the occurrence of Si(2)H(5) on the surface thus providing evidence for silicon-hydrogen bond dissociation during disilane adsorption. Specifically, we find that an experimentally observed vibrational peak at 2150 cm(-1) that has generally been attributed to the trihydride SiH(3) is more likely to be due to Si(2)H(5). Our results also clear up the assignment of two peaks for monohydride species adsorbed at the edge of a growing terrace, and a peak for the dihydride species adsorbed in the interdimer configuration.