The real significance of three-dimensional diffusion simulation in present process technology lays in the influence of the boundary conditions. The physics at the silicon/silicon-dioxide and silicon/nitride interface has a strong impact on the dynamics of point defects and thus on the diffusion of dopants [33,28,20].
It has been established [34,21,20] that the interactions at these interfaces can be described by a simple phenomenological model, given by (3.67) and (3.68), which still allows the appropriate description of two basic effects: silicon boundary as perfect reflector for point defect and silicon boundary as perfect point defect sink. Some researchers argue the complete dominance of surface processes over bulk processes [34].
In order to illustrate three-dimensional effects and capabilities of the numerical schemes presented in Sections 3.8.3 and 3.8.5 we use a setup of process steps preeceding the diffusion process step as follows. For performing the three-dimensional process simulation flow the topography simulator TOPO3D [44,45] has been used. Deposition and etching process steps have been carried out with the empirical model of TOPO3D.
The following pictures show as example a sequence from the three-dimensional simulation of typical CMOS technology process flow.
Before source/drain formation, a deposition of the conformal Si
N
layer is carried out (Figure 3.8).
The thickness of this layer will determine the thickness of the sidewall spacer and it is used to optimize device characteristics [5].
Subsequentially, the deposited SiN layer is anisotropically etched (Figure 3.9), leaving a sidewall spacer along the edge of the polysilicon.
In the prepared structure an initial arsenic profile is introduced with the multi-dimensional ion implantation simulator MCIMPL-II [39,40].
Implanted is the dosis of
cm
with the energy of 
keV.
For comparison reason the simulation is performed simultaneously using the Mulvaney-Richardson model (Section 3.4.2) with the surface recombination model (Section 3.5.4) and without it. To obtain high nonequilibrium dynamics of the point defects the one-plus model is applied.
The simulation result for the interstitial concentration without surface reaction model is presented in Figure 3.10a and with the surface reaction model in Figure 3.10b. In both cases the simulations have been carried out at 950 
C for the duration of approximately 1 s.
Figure 3.10b shows a significant lowering of the interstitial profile not only directly in the vicinity of the silicon/silicon-dioxide interface but also deeper in the bulk. The silicon surface acts as sink of the interstitials, trying to reduce the concentration level from that given by the one-plus effect to the equilibrium level.
Since the silicon/silicon-dioxide interface has an explicitely three-dimensional shape, the resulting interstitial profile will also show three-dimensional spreading.
Since the interstitals are the main vehicles of enhanced arsenic diffusion, the dwindling of interstitals reduces the depth of the diffused arsenic profile. This can be clearly seen from the Figure 3.11 (no surface recombination) and Figure 3.12 (with surface recombination). The good example of the experimental results confirming the strong influence of the silicon/silicon-dioxide interface on the transient enhanced diffusion can be found in [46].
|
|
|
].![$\textstyle \parbox{5cm}{
\vspace*{2.2cm}
$1.2 \cdot 10^{18}$\ [6mm]
$9.2 \c...
...t 10^{17}$\ [6mm]
$3.1 \cdot 10^{17}$\ [6mm]
$9.5 \cdot 10^{15}$\ [6mm]
}$](img715.png)
![$\textstyle \parbox{5cm}{
\vspace*{2.2cm}
$3.0 \cdot 10^{19}$\ [6mm]
$2.4 \c...
...t 10^{19}$\ [6mm]
$5.9 \cdot 10^{18}$\ [6mm]
$2.4 \cdot 10^{11}$\ [6mm]
}$](img718.png)
![$\textstyle \parbox{5cm}{
\vspace*{2.2cm}
$4.1 \cdot 10^{19}$\ [6mm]
$3.3 \c...
...t 10^{19}$\ [6mm]
$8.2 \cdot 10^{18}$\ [6mm]
$2.1 \cdot 10^{12}$\ [6mm]
}$](img720.png)