5.1 Polysilicon Emitter

  Dopant diffusion in polysilicon material exhibits strong geometry effects which must be traced by two-dimensional simulations. Due to the lack of two-dimensional measurement techniques we have to compare one-dimensional cross sections of two-dimensional simulations with SIMS measurements. Nevertheless, the two-dimensional dopant distribution is important for subsequent device simulations to determine the electrical characteristics of the device. To examine the applicability of our polysilicon diffusion model we perform an outdiffusion simulation for a bipolar device. The initial device structure prior to annealing is given by Figure 5.1-1, consisting of a wide single polysilicon emitter.

  
Figure 5.1-1: Initial simulation structures for a bipolar transistor with a wide single poly emitter.

We started up with a n-type wafer, where the collector was already fabricated with a constant doping level of . The intrinsic base is then implanted with boron at 35keV energy with a dose of into the silicon substrate. After fabrication of the oxide isolation and a 1nm thick interfacial oxide, a polysilicon layer was deposited at and doped with arsenic. The orientation of the polysilicon main grain axis obtained from the previous deposition simulation as well as the stress distribution in the polysilicon layer are given in Figure 5.1-2 and Figure 5.1-3, respectively.

  
Figure 5.1-2: Orientation of the grain main axis obtained from the deposition simulation.

The initial dopant concentration is shown in Figure 5.1-4. During 15s RTA annealing at , outdiffusion of the arsenic dopants from the polysilicon layer into the underlying substrate took place. The final total arsenic distribution in the polysilicon layer as well as in the substrate is depicted in Figure 5.1-5. The total concentration suggests that the polysilicon grain boundaries are filled up with dopants to their areal limit . Investigating the arsenic grain boundary concentration (see Fig. 5.1-6), we find lower grain boundary concentrations in regions where dopant dependent grain growth took place. The grain size determines the maximum grain boundary concentration during annealing.

The dopants in the grain bulk regions are totally activated, which leads to a flat profile for the arsenic grain interior concentration (see Fig. 5.1-7). Additionally, we give the lateral size of the polysilicon grains in Figure 5.1-8. Due to the high interface doping concentration a significantly increased growth rate is observed. In regions with high stress levels the growth rate is retarded.

    
Figure 5.1-3: Stress distribution within the polysilicon layer according to simulation geometry.
Figure 5.1-4: Initial dopant profiles after the polysilicon layer deposition and implantation. Arsenic is shown in the polysilicon layer, where the boron base doping is given in substrate.

    
Figure 5.1-5: Resulting total arsenic distribution after 15s RTA annealing at . In the polysilicon layer the grain boundary and grain interior concentration was added.
Figure 5.1-6: Arsenic concentration in the polysilicon grain boundaries. At regions with high grain growth rates the arsenic grain boundary concentration is lowered.

  
Figure 5.1-7: Arsenic concentration in the polysilicon grains. As all dopants are activated the distribution is nearly flat in the whole polysilicon layer.

  
Figure 5.1-8: Vertical size and orientation of the polysilicon grains. Enhanced grain growth starts from the polysilicon-monosilicon interface.

The net doping concentration in the silicon substrate is given by Figure 5.1-9, where Figure 5.1-10 gives the one-dimensional dopant profiles of a cross-section in the mid-emitter region. The emitter profile is high enough to achieve sufficient current gain and to keep the emitter resistance low. The base width is 120nm and the Gummel number is .

  
Figure 5.1-9: Net doping profile in the active area of a NPN-transitor.

  
Figure 5.1-10: One-dimensional cross-section of the dopant profiles in the mid-emitter.