3.4 Process Simulation

Process simulation models every process step of a silicon wafer from start of processing until electrical wafer acceptance test. The different process steps possible were outlined in Section 2.5. As mentioned in the previous section, the level of detail of input parameters to the simulator is heavily depending on the complexity of the physical model implemented in the process simulator. These input parameters are varying, depending on the type of process step. An example of such parameters is given in Table 3.1.
The process simulation flow consists of a sequence of single commands for the process simulator, naming the type of the process steps (diffusion, etching, etc.) and its process conditions in terms of parameters (time, temperature, pressure, etc.). A detailed review of the syntax of such a process flow is given in Chapter 5.
A second class of parameters are the model parameters which are the result of a careful calibration of the models for a certain range of input parameters (like e.g.temperature). These parameters describe the underlying physics of the process simulator model. Figure 3.6 shows this situation.

Figure 3.6: Parameter classes for process simulation

There are several types of applications for process simulation in the semiconductor manufacturing industry and research.

Figure 3.7: Types of process simulation applications

(a)

(b)

(c)

First, (especially during the process flow development phase) single process steps are simulated and optimized for a certain targeted feature like oxide distance or well depth. A good example of such optimizations is the transfer of furnace recipes between different wafer sizes given as a detailed example in Section 6.1. In these situations one is mainly interested in optimizing the main step of a diffusion recipe to gain identical thermal budgets for both recipes thus obtaining the same diffusion profile on different wafer topologies. The algorithm is shown in Figure 3.7(a).
Second, during integration of additional modules in an already available process flow, process simulation is used to identify key influences of the module on the overall process flow. The term module means a combination of process steps which form a unit to make a certain feature of the process technology like the isolation between semiconductor devices (LOCOS for older technologies and shallow trench isolation for newer technologies). A good example is, e.g., added thermal budget because of additional diffusion steps inside the new module. This additional thermal budget must be minimized to gain the same features of process steps preceding the module as given by the process flow without the additional module (e.g. the doping profile of a preceding threshold adjust implant is broadened by the additional thermal budget, thus the implant dose has to be increased and/or the implant energy lowered to maintain the same surface concentration of the threshold adjust implant). The algorithm is shown in Figure 3.7(b).
Third, an already installed process flow is simulated to generate the electrical parameters of semiconductor devices by a process simulation followed by a subsequent device simulation. This type is mainly used for optimizing the process technology in terms of parametric performance. Furthermore, in terms of process complexity (cost) simulations are carried out to optimize an already implemented process technology further. In addition the statistical sensitivity of device parameters (like threshold voltage or on-resistance) on certain process flow parameters like implant energy or dose of a certain implant can be calculated. The algorithm for this strategy is shown in Figure 3.7(c)