5.4.1 Threshold Voltage Adjust Implantation of an NMOS-Transistor

Fig. 5.24 shows the input structure for the ion implantation process step. It is one half of an NMOS-Transistor which is cut through the gate. The crystalline silicon substrate is partly covered with a thick silicon dioxide layer forming the isolation and a thin silicon dioxide layer serving as a scattering oxide. The isolation is formed by a local oxidation of silicon (LOCOS) process.

Figure 5.24: Structure of an 0.6 $\mu$m NMOS-transistor before the threshold voltage adjust implantation.

Figure 5.25: Simulated boron (top) and fluorine (bottom) profile resulting from a threshold voltage adjust implantation into a 0.6 $\mu m$ NMOS-transistor with BF$_2$ ions with an energy of 50 keV and a dose of $3{\cdot }10^{12}$cm$^{-2}$. The figure shows an outline of the transistor structure and the doping profile within three cuts through the gate and source/drain region.

The implantation is performed with BF$_2$ ions with an energy of 50 keV and a dose of $3{\cdot }10^{12}$cm$^{-2}$. The ion beam was tilted by 7 $\,^\circ\;$and rotated by 0 $\,^\circ\;$ . By this implantation a boron concentration up to $2{\cdot}10^{18}$cm$^{-3}$ is introduced in the silicon substrate, which has a positive background doping of $2{\cdot}10^{15}$cm$^{-3}$, below the gate region (narrow region below the scattering oxide) and also below the source/drain region (large rectangular region below the scattering oxide) as shown in the top figure of Fig. 5.25.

Just the doping below the gate has a significant influence on the final device behavior, while the source/drain region is additionally doped by a succeeding ion implantation steps with a higher implantation dose. The effect of this implantation is to adjust the threshold voltage of the transistor. The threshold voltage is defined as the gate voltage above which the transistor becomes conductive due to an inversion of a thin layer below the gate. The voltage which is necessary to create an inversion layer strongly depends on the original doping concentration, which is adjusted by this implantation. Threshold voltage adjust implantations are always performed with low doses, because just slight modification of the gate concentration are sufficient for the adjustment. Typical doses are $4{\cdot}10^{11}$cm$^{-2}$ to $6{\cdot}10^{12}$cm$^{-2}$ ([91]).

The simulation was performed using the molecular method (Sec. 4.5.2). Therefore, the impurity distributions of the boron atoms and of the fluorine atoms have been calculated. Fig. 5.25 shows the distribution of the boron atoms (top) and of the fluorine atoms (bottom) in the transistor. The doping profiles are visualized within three cuts through the active area of the transistor in common with an outline of the transistor structure. 2300000 distinct ions where simulated on a DEC-600 workstation with 333 MHz CPU clock frequency. The simulation took approximately 5 hours of CPU time.

As expected there is not much difference between the boron and the fluorine distribution, except that the fluorine concentration is twice the boron concentration due to the larger number of fluorine atoms in the BF$_2$ molecule. Therefore an ion implantation with BF$_2$ ions can also be simulated quite accurately with the simplified molecular method as already mentioned in Sec. 4.5.2. If the same simulation is performed with the simplified molecular method (Sec. 4.5.2) the simulation time reduces to 3 hours.

Alternatively this simulation can be performed with the analytical method (Sec. 3.1) or if a higher accuracy is required with the point response interface method (Sec. 3.2). The analytical method requires a simulation time of approximately 30 minutes while the point response interface method requires a simulation time of approximately 2 hours. By using the point response interface method the simulation time is mainly determined by the time required for the calculation of the point response function.