SIC FORMATION BY REACTION OF SI(001) WITH ACETYLENE - ELECTRONIC-STRUCTURE AND GROWTH MODE

The carbonization process of a single domain 2X1-reconstructed Si(001) vicinal surface (5 degrees off axis from [001] in the [(1) over bar 1 0] direction) in acetylene has been studied by combining in situ surfa ce science techniques (x-ray photoemission spectroscopy, x-ray photoel ectron diffraction, reflection-electron energy loss spectroscopy, low- energy electron diffraction) and ex situ analytical techniques (C-12 a nd H-2 dosing by nuclear reaction analysis, scanning electron microsco py, and reflection high-energy electron diffraction). It is found that at a growth temperature of about 820 degrees C a variety of growth me chanisms can be observed, particularly during the first step of carbon ization. An analysis of C 1s and Si 2p core-level shifts and of the re spective intensities of them, combined with the examination of photoel ectron diffraction curves, gives evidence for a penetration of C atoms into the silicon substrate, to form a nonstoichiometric compound. Con temporaneously 3C-SiC nuclei form, aligned with respect to the substra te. Then a quasicontinuous 3C-SiC film grows heteroepitaxially (''cube on cube'' unstrained growth) on the substrate up to a thickness of si milar to 40 Angstrom. C 1s and Si 2p photoelectron diffraction pattern s, compared with calculated ones, show that the single domain initial surface does not necessarily force a preferential alignment of one of the two inequivalent SiC{110} planes with respect to the (110) Si plan e. Consequently, such vicinal Si(001) surfaces are not necessarily tem plates, as often reported in the literature. for the growth of crystal line films free of antiphase boundary domains. Finally, we have observ ed that an imperfect coalescence of 3C-SiC nuclei leaves easy paths fo r Si out migration from the substrate and SiC polycrystalline growth, even at a temperature as low as 820 degrees C. The current models of S i(001) carbonization are examined and compared to our experimental fin dings. Especially for the very beginning of carbide formation, a unifi ed picture is lacking, as the role played by the steps and terraces of the initial surface remains unclear.