3.2.2 High Concentration Effects (Models DIFSCL and DIFDCL)



next up previous contents
Next: Dynamic Clustering (Model Up: 3.2 Model Library Previous: Model Parameters

3.2.2 High Concentration Effects (Models DIFSCL and DIFDCL)

 

During VLSI technology processes, mainly for shallow junction formation, dopant concentrations higher than solid solubility and off equilibrium conditions are generally employed. At high dopant concentrations not all dopant atoms are ionized due to clustering and precipitation. The total concentration of an impurity will be made up by a mobile (active) part and an immobile (clustered) part , (). This cluster formation causes a drastic reduction of the diffusion coefficient at high concentrations because the gradient of the active atoms almost vanishes and the clustered part of the dopants does not diffuse (see Figure 3.2-1).

 

For boron and antimony the inactive dopants form relatively large precipitates, whereas the inactive arsenic and phosphorus atoms form very small coherent precipitates or clusters [Sol90]. An impurity cluster is composed of impurity atoms and electrons. The cluster formation process (3.2-10) does not occur instantaneously, several hours at and a few minutes at are necessary to attain equilibrium conditions [Sol90].

 

Commonly used process simulations are inadequate to correctly predict the dopant redistribution in supersaturated conditions. SUPREM III [Ho83b] and PREDICT [Fai88], even though assuming a maximum electrically active solubility, consider all dopants mobile during thermal treatment. On the contrary, ICECREM [Pic90] hypothesizes the instantaneous aggregation of dopant atoms to form immobile clusters.

In PROMIS' model library we supply two models for clustering effects, a kinetic (dynamic) clustering model DIFDCL and an equilibrium (static) clustering model DIFSCL.





Martin Stiftinger
Wed Oct 19 13:03:34 MET 1994