NIKMI zyxwvutsrq B Nuclear Instruments and Methods in Physics Research B 85 (1994) 248-254 North-Holland Beam Interactions with Materials&Atoms IBA study of the growth mechanisms the effect of wafer cleaning * of very thin silicon oxide films: F.C. Stedile +, I.J.R. Baumvol ++, J.-J. Ganem, S. Rigo, I. Trimaille, G. Battistig +++ zyxwvutsrqponm Groupe de Physique des Solides, Universitb de Paris 7 et 6 - UR417 du CNRS, Tour 23, 2, Place Jussieu, 75251 Paris Cedex 05, France W.H. Schulte, H.W. Becker Institut fiir Experimentalphysik III, Ruhr-Universitiit Bochum, Bochum, Germany The growth mechanisms of very thin silicon oxide films formed during rapid thermal oxidation were studied using ion beam analysis and 180 isotopic tracing methods. In this paper we report on the effects of different cleaning procedures of silicon wafers prior to oxidation in dry oxygen (1602 followed by “0,) on the growth mechanisms and kinetics. Typical oxide thicknesses ranging from 0.2 to 10 nm were studied. The I80 and 160 isotopic profiles were determined by ion beam analysis methods, namely: the ‘sO(p, (y)15N narrow resonance at 151 keV, and the “O(p, c~)“N and the 160(d p)170 reactions associated with step-by-step , chemical dissolution. The profiles could be related to current theories on the initial stages of thermal growth of silicon oxide layers allowing us to draw some conclusions regarding the role of surface cleaning of the silicon wafers on the formation of silicon fragments in the volume of the very thin oxide layer. The influence of rapid thermal processing parameters like temperature, time and oxygen partial pressure on the growth mechanisms were also studied and discussed here. 1. Introduction As silicon oxidation is one of the most common steps in the fabrication process of electronic devices it is not surprising to see a great effort spent continu- ously, during the last decades, to understand the growth kinetics of silicon oxide films on silicon wafers. The thermal oxidation of silicon is usually discussed under the light of the Deal and Grove model [l], which assumes that the oxidizing species CO,, H,O) are dis- solved in the silica network in interstitial positions and migrate to the SiO,/Si interface where they oxidize the silicon substrate. However, it was verified that this model cannot predict the behavior of the oxidizing + Corresponding author. Present address: Instituto de Quimica, UFRGS, Av. Bento Gonqalves, 9500, Porto Alegre RS, 91540-000 Brazil.Phone + 55 51339 1355, fax +55 513363699. ++ Permanent address: Instituto de fisica, UFRGS, Port0 Alegre, Brazil. +++ Permanent address: KFKI Research Institute for Mate- rials Science, H-1525 Budapest, Hungary. * Work supported in part by the French CNRS, GDR86, by the Brasilian CNPq and by the Hungarian OTKA grants No. 1829 and 3265. species in the case of dry oxidation of silicon when the resulting silica film is thinner than 30 nm [2]. For this specific case, there is still no consensus about the growth mechanism. One of the most suitable methods to study this field is the 180 isotope tracer technique, which is well described in ref. [2]. For oxide films thicker than 15 nm grown by rapid thermal oxidation (RTO) (see Fig. 1) or by classical oxidation [3], the isotopic tracing experiments (oxida- tion in 1602 followed by an oxidation in “0,) lead to the formation of two regions rich in “0 in the silicon oxide film separated by a region of Si1602. One of these 180 rich regions is at the SiO,/Si interface, as predicted by Deal and Grove, and the other one is at the external surface. This last region is attributed to mobile oxygen defects which allow a step-by-step trans- port of oxygen. For thinner films grown by RTO, however, some “0 is also present in the bulk of the oxide. It was suggested that the “0 found in the volume of the oxide corresponds to Si grains or frag- ments inside the Si’60, layer which were left non- oxidized after the first thermal treatment of the wafer in 1602 but were oxidized in the subsequent treatment in “0, [4]. The Si fragments were previously observed by TEM [5] and RHEED [6]. In ref. [4] a model to 0168-583X/94/$07.00 0 1994 - Elsevier Science B.V. All rights reserved SSDI 0168-583X(93)E0460-X