Single-Source CVD of 3C-SiC Films in a LPCVD Reactor II. Reactor Modeling and Chemical Kinetics Gianluca Valente, a Muthu B. J. Wijesundara, b Roya Maboudian, b and Carlo Carraro z a Dipartimento di Chimica, Ingegneria Chimica e Materiali-G. Natta, Politecnico di Milano, I-20131 Milan, Italy b Department of Chemical Engineering, University of California, Berkeley, California 94720 The deposition of 3C-SiC films on Si100 wafers from 1,3-disilabutane precursor 1,3-DSB molecule utilizing a conventional low pressure chemical vapor deposition LPCVD system is studied theoretically. The LPCVD reactor is modeled as a 1D system with axial dispersion. The reaction path of the 1,3-DSB decomposition is studied through quantum chemical methods due to the lack of kinetic information on this molecule. First, the transition state for the decomposition reaction is calculated, then the fall-off regime of the reaction is determined using the Rice-Ramsperger-Kassel-Marcus theory. The surface kinetics is taken-mainly from literature data. The kinetic scheme obtained in this manner is first embedded into a 0D model to evaluate the most important kinetic processes, then a simplified kinetics is adopted in the 1D model. Finally, the calculated growth rates are compared with experimental data taken at different temperatures. The agreement between calculated and experimental data shows the importance of the gas-phase decomposition reactions during the SiC deposition even in a low-pressure process. © 2004 The Electrochemical Society. DOI: 10.1149/1.1646142 All rights reserved. Manuscript received December 16, 2002. Available electronically February 9, 2004. The silicon carbide deposition described in the preceding paper 1 is studied here from a theoretical point of view. Silicon carbide is grown in a commercial low pressure chemical vapor deposition LPCVD reactor on Si100 wafers with a single precursor 1,3- disilabutane, CH 3 SiH 2 CH 2 SiH 3 1,3-DSB. The main theoretical study on silicon carbide CVD has been performed by Allendorf and Kee. 2 They focused their attention on the high-temperature pro- cesses in order to obtain epitaxial silicon carbide, by feeding two different precursors molecules, SiH 4 and C 3 H 8 , as silicon and car- bon sources, respectively. The authors investigated both gas-phase and surface chemistry, including SiH 4 and C 3 H 8 pyrolysis, radical adsorption, and hydrogen desorption. The same system was also studied by Raffy et al., 3 who investigated the kinetic pathways lead- ing to the formation of methylsilane (CH 3 SiH 3 ) in the gas phase through ab initio calculations. In contrast, theoretical investigations of single precursors in gen- eral, and 1,3-DSB in particular, are missing. This may be due in part to the fact that no single molecule has yet emerged as the standard single precursor. Some theoretical and experimental studies have focused on methyltrichlorosilane (CH 3 SiCl 3 ), 4,5 and on methylsi- lane (CH 3 SiH 3 ). 6 Methylsilane, in particular, is relevant to the present work because of the chemical analogy with 1,3-DSB. With macro/microcavity experiments, Oshita found the presence of two different precursors in the gas phase. 6 It was concluded that a gas- phase reaction likely takes place, leading to the formation of a spe- cies with high sticking coefficient. The decomposition of CH 3 SiH 3 to CH 3 SiH was suggested as the most probable reaction. In the present work, the 1,3-DSB gas-phase reaction pathway is calculated with quantum chemical calculations. A kinetic scheme is obtained which is then used with a 0D model, simplified and, finally, embedded in a 1D reactor model to calculate the growth rate pro- files. The most important gas-phase reaction is found to be the de- composition of 1,3-DSB to give CH 3 SiH 2 CH 2 SiH species. The ef- fect of this reaction, which is analogous to that proposed by Oshita et al. for CH 3 SiH 3 , is to produce a second growth precursor in addition to 1,3-DSB. The presence of two precursors in the silicon carbide growth explains the particularly sharp growth rate profiles that are observed experimentally at high temperature. 1 Reactor Model The reactor is already described in part I of this work. 1 Briefly, it is a horizontally oriented hot-wall tubular reactor TekVac, CVD- 300-M; the quartz tube has an inner diameter of 75 mm and is 600 mm long with a 450 mm long hot-wall zone with temperature uni- formity of 1°C. The reactor base pressure is less than 10 -7 Torr using an 80 L/s turbomolecular pump. The precursor is 1,3-DSB Gelest Inc., 95% purity, further purified by freeze-pump-thaw cycles using liquid N 2 before introduction into the reactor via a mass flow controller MKS SDS-1640. The low-pressure CVD reactor is modeled with a one- dimensional, axial dispersion model. Thus, the species concentration is considered as a function of the axial coordinate, assuming that its value is constant along the transverse reactor section. This assump- tion is valid because of the low operating pressure 50 mTorr; at such low pressures, the mean-free path of the molecules in the gas phase becomes comparable with the reactor diameter and, thus, the diffusion resistance becomes very low. According to these assump- tions, the global material balance and the material balance for each gas-phase species can be written as d  v C T v i  dz - d dz  C T D i,mix dy i dz  = R i gas + R i sur i = 1, ..., NCG-1 1 d  v C T  dz =  =1 NCG R i gas +  I =1 NCG R i sur 2 where v is the mean gas velocity, C T is the total gas concentration evaluated using the ideal gas law P /( RT ) with R being the gas constant, y i is the mole fraction of the ith species, D i,mix is the diffusion coefficient of the ith species in the gas mixture, and R i gas and R i sur are the production rate of the ith species in the gas phase and on the surface, respectively. NCG is the number of species in the gas phase. The standard Dankwerts boundary conditions are im- posed on the reactor boundaries for the mass balance of each spe- cies, while the inlet gas velocity is imposed for the global material balance equation  v F y i F = v  0  y i  0  - D i,mix dy i dz  0 + v  0  = v F 3 z E-mail: carraro@cchem.berkeley.edu Journal of The Electrochemical Society, 151 3 C215-C219 2004 0013-4651/2004/1513/C215/5/$7.00 © The Electrochemical Society, Inc. C215 ) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 54.196.252.114 Downloaded on 2016-10-24 to IP