Ab initio modeling study of boron diusion in silicon W. Windl a, * , R. Stumpf, b , X.-Y. Liu b , M.P. Masquelier b a Computational Materials, Motorola, Inc., Austin TX, USA b Computational Materials, Motorola, Inc., Los Alamos NM, USA Accepted 18 April 2001 Abstract We present investigations of boron diusion in crystalline silicon using ab initio calculations W. Windl et al., Phys. Rev. Lett. 83 1999) 4345). Based on these results, a new mechanism for B diusion mediated by Si self-interstitials was proposed. Rather than kick-out of B into a mobile channel-interstitial, one- or two-step diusion mechanisms have been found for the dierent charge states. The predicted activation energy of 3.5±3.8 eV, migration barrier of 0.4±0.7 eV, and diusion-length exponent of 0:6to 0:2 eV are in excellent agreement with experiment. We also present results of ab initio calculations for the structure and energetics of boron-interstitial clusters in Si. We show how these ®rst-principles results can be used to create a physical B diusion model within a continuum simulator which has strongly enhanced predictive power in comparison to traditional diusion models. Ó 2001 Elsevier Science B.V. All rights reserved. Keywords: Boron diusion; Ab initio process modeling 1. Introduction Ion implantation is currently the method of choice for introducing dopants such as boron into silicon [1]. However, energetic ions cause damage in the host material and create a supersaturation of defects in Si, which impair the device performance. Annealing following the implant is used to heal the implant damage, while activating the dopant at- oms electrically at the same time. The implant- anneal cycle can cause on the one hand excessive transient enhanced diusion TED) of the im- planted B, and on the other hand the formation of B precipitates which immobilize and de-activate the B atoms well below the solid solubility limit [2,3]. From the observation of the trapping of in- terstitials Is) by these precipitates [4], it was con- cluded that they consist of B±I clusters BICs). To understand both processes ± diusion including TED) and clustering ± is the basis for successful physical process simulation. Both experimental observations and theoretical calculations indicate that diusion of common dopants in Si is mediated by Is or vacancies Vs) [5,6]. Experiments based on the injection of point defects into a B doped Si substrate strongly sug- gest that B diusion in Si is dominated to a degree of more than 98% by an I mechanism [7]. While experiments are able to ®nd the total diusion coecient and the in¯uence of point-defects on it, they generally cannot determine the microscopic diusion mechanism [8]. However, to accurately model dopant diusion under conditions in¯u- enced by externally created defects, dopant pairing Computational Materials Science 21 2001) 496±504 www.elsevier.com/locate/commatsci * Corresponding author. E-mail address: wolfgang.windl@motorola.com W. Windl). 0927-0256/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII:S0927-025601)00197-5