Anomalous Transient Tail Diffusion of Boron in Silicon: Kinetic Modeling of Diffusion and Cluster Formation
- PDF / 1,096,675 Bytes
- 4 Pages / 420.48 x 639 pts Page_size
- 23 Downloads / 269 Views
ANOMALOUS TRANSIENT TAIL DIFFUSION OF BORON IN SILICON: KINETIC MODELING OF DIFFUSION AND CLUSTER FORMATION
N.E.B. COWERN*, H.F.F. JOS**, K.T.F. JANSSEN*** AND A.J.H. WACHTERS*
Philips Centre for Manufacturing Technology, P.O. Box 218, 5600 MD Eindhoven, The Netherlands Philips Components, Gerstweg 2, 6534 AE Nijmegen, The Netherlands * Philips Research Laboratories, P.O. Box 80000, 5600 JA Eindhoven, The Netherlands
*
ABSTRACT A kinetic modeling approach is used together with experimental studies to gain insight into the processes taking place during anomalous transient diffusion. A physical process analogous to B clustering is found to play an important role, even at B concentrations well below the solid solubility limit.
INTRODUCTION The mechanisms of implantation-induced transient diffusion have been the subject of study and speculation for more than 15 years, but considerable uncertainty still remains. From the technological point of view, this limits the effectiveness of computer-aided process design tools, as the vertical dimensions of the simulated devices shrink to the same order as the transient diffusion length. From a scientific perspective, the
challenge is to identify relationships between the complex diffusion behaviour that is observed and basic interactions that might control this behaviour, such as kickout of dopant atoms and nucleation and growth
of precipitates. Transient diffusion after B implantation into silicon provides a typical example of the problem. Over a rather wide range of implantation doses one observes a rapid diffusion in the tail of the implant profile during annealing. At the same time, relatively little diffusion occurs near the peak of the B profile. Several ideas [1-4] have been put forward to account for this anomalous diffusion behaviour, which is also seen in other dopants such as P [2] and Ga [5]. A number of workers have pointed out that anomalous diffusion profiles can arise when only part of the implanted dopant is assumed to be mobile. Hofker et al [1] were able to account for observed B diffusion profiles with peak concentrations above the solid solubility limit in terms of a 2-component model where a large fraction of the implanted dopant was assumed to be immobile. This fraction was attributed to B precipitates. Morehead and Hodgson [2] analysed anomalous P transient diffusion with a 2-component model and suggested that their mobile component should be associated with dopant-defect pairs. Finally, Michel [3] proposed a 2-component model in which the B is initially immobile and electrically inactive, and decays exponentially with time into a mobile, electrically active component. The model of Ref. 2 closely resembles the physical mechanism responsible for normal, steady-state dopant diffusion, if we identify the static component as substitutional atoms. In this case, the only difference between transient and normal diffusion is the extreme supersaturation of the mobile species during transient diffusion. In the models of Refs. 1 and 3, electrically inactive d
Data Loading...
