Low Energy Ion Implantation in Si

Ion implantation is the key doping process in Si integrated circuit technology. As the scale of CMOS devices shrinks, low energy ion implantation is required to reduce source/drain extensions in order to avoid short channel effects. B+ ion implantation is the most useful p-type doping process in Si and it is vital to determine as-implanted and post-anneal impurity distributions with the highest possible accuracy.

Secondary ion mass spectrometry (SIMS) is generally employed for this task but, while providing the basic dopant profile, it invariably indicates the presence of a narrow surface concentration peak. The origin of the latter has been uncertain, but it has been suggested that it may be due to redistribution of subsurface B during initial stages of Si surface sputtering.

However, in order to clarify the situation, the present work has employed EFI in the FEGTEM as the basis for a high resolution study of the B distribution in such implanted Si. A typical EFI-derived impurity distribution in Si implanted with 1keV B+ ions to a dose of 5x1015 ions/cm2 is shown in Fig. 1.

Figure 1: Comparison of as-implanted B distributions showing EFI profile, TRIM simulation profile and SIMS profile.

It is immediately clear that the distribution determined is smooth and does not display the surface peak characteristic of SIMS. Indeed, the distribution measured by EFI is similar to that theoretically predicted by the TRIM simulation method (also shown): Rp is the predicted ion range. Therefore, the surface B concentration peaks invariably given by SIMS analyses may well need to be re-evaluated.

The EFI work has also been extended to study such implanted layers after annealing. It has then been directly shown that the implanted B aggregates into clusters which have dimensions of a few nanometres and which exhibit a disordered internal structure (see Fig. 2).

Figure 2: a) B map obtained from 900°C annealed sample showing clustered impurity and b) 004 strain-contrast image of same area showing disordered zon

The way in which B high temperature diffusion takes place will be modified by the presence of these clusters and this information is of substantial importance for use in process simulation tools employed by device manufacturers.