In the past the final dopant profile was determined by diffusion over relatively large distances. With decreasing feature size, the amount of diffusion was reduced and hence the initial dopant distribution after implantation and the anomalies during the initial stages of diffusion became progressively important. These anomalies consist of a surprisingly high transient rate of diffusion caused by interaction with point defects originating from the earlier implantation. The underlying microscopic processes are still incompletely understood, making it difficult to write down the equations governing the process. Therefore, it is even more important that the simulation software supports the input of a new model for its speedy evaluation. This is already true for one-dimensional problems. For deep submicron processes tight control over the lateral diffusion becomes essential, hence full two-dimensional diffusion simulation is indispensable. In addition, the need for swift adaption of simulation capabilities to changing technological requirements asks for simulation software which allows easy implementation of new or updated models which can be handed out again to a device or process engineer.
Full modeling of ion implantation requires a detailed description of all scattering processes. This is used to generate the spatial distribution of ions implanted into an arbitrary structure, together with the spatial distribution of atoms displaced by the implantation process. The latter are significant for instance for modeling subsequent diffusion steps. The implantation simulation can be carried out either by numerical solution of the Boltzmann transport equation for the ions and recoils, or by the equivalent Monte Carlo method. From the physical point of view, ion implantation is one of the best understood processes within silicon technology. Computational limitations such as handling implantation into complex geometrical structures are the main constraints on accuracy in this case.
Approximate solutions applicable to many situations can be obtained by solving for the spatial distributions formed by a point source and using a convolution method to simulate implantation into real structures. This approach provides a substantial reduction in computational requirements which is essential for routine simulations.