8.6 Example: A Parasitic MOSFET



next up previous contents
Next: 8.6.1 Data Exchange Aspects Up: 8 Example: Integration of Previous: 8.5 Example: A Microstrip

8.6 Example: A Parasitic MOSFET

 

  
Figure 8.10: Geometry of the parasitic MOSFET

  
Figure 8.11: Geometry with extracted -junction

As a second example a parasitic MOSFET resulting from the isolating oxide of a LOCOS process is presented. Fig. 8.10 shows one half of the geometry of this parasitic MOSFET. The top of the highly nonplanar oxide is covered by the gate metalization for adjacent MOS transistors. This metalization induces MOSFET-like behavior of the structure in question, although it is originally intended as an isolation between the devices.

  
Figure 8.12: Grid generated by MESHCP

Below the oxide the silicon is lowly -doped, and on the rear it is covered by the bulk contact. On the left, the -doped region with the source contact gives rise to a -junction which can be seen in Fig. 8.11 as extracted after the MESHCP geometry preprocessing step. Fig. 8.12 shows the initial grid generated by MESHCP, on which the capacitance calculation is performed.

  
Figure 8.13: Gate-Bulk capacitance versus gate voltage

VLSICP is run in a loop applying gate voltages from to to create the C-V characteristics of the parasitic MOSFET in various operating conditions. Fig. 8.13 depicts the results of this calculation, showing a typical MOSFET input capacitance curve [Aror93][Sze81]. In the first section, ranging up to , the gate-bulk capacitance is steadily decreasing because of the increasing width of the depletion layer under the gate. However, as the onset of inversion creates a substantial amount of electrons at the oxide-semiconductor interface, this channel charge forms a counter electrode to the gate capacitance. As a consequence, the capacitance increases and finally reaches a saturation value when the device operates in the strong inversion regime (at gate voltages of and above). This saturation value is the oxide capacitance given as

Due to the nonplanar gate oxide, inversion will start under the left edge of the oxide close to the source contact, where the oxide is thin. However, the presence of charge carriers in this region will already increase the capacitance, which is the case around approximately gate voltage. As the gate voltage is increased, the inversion layer will more and more extend to the right end the symmetric axis). The threshold voltage will be reached, when inversion occurs under the whole gate oxide. Therefore the threshold voltage can just roughly be estimated from the capacitance curve. A value of approximately is obtained, depending on the location under the gate, which compares well to the results of a device simulation [Fisc94]. Since this structure was designed as an isolating oxide, we can see that the design goal was not met, because at gate voltages of the structure would no longer isolate.

  
Figure 8.14: Potential distribution at gate voltage

  
Figure 8.15: Potential distribution on vector product grid

Fig. 8.14 and Fig. 8.15 demonstrate VLSICAP's ability to write output on its native six-node triangular grid as well as on a tensor product grid respectively. Both show the electrostatic potential at a gate voltage of . The kink at the oxide-semiconductor interface is due to the discontinuity of the permittivity.





next up previous contents
Next: 8.6.1 Data Exchange Aspects Up: 8 Example: Integration of Previous: 8.5 Example: A Microstrip



Martin Stiftinger
Tue Nov 29 19:41:50 MET 1994