Q: Can the workfunction of the MOS polysilicon gate contact be calculated by ATLAS based on the doping? Can poly depletion effects be simulated in ATLAS?

A: The polysilicon gate contact in MOS devices can be simulated in two distinct ways using ATLAS. These correspond to treating the polysilicon region as

  • A metal-like equipotential region with a specified workfunction.
  • A semiconductor region with a potential defined by the doping level.
Most commonly the former approach is adopted. The polysilicon region acting as the gate is defined as an electrode in ATHENA. The ELECTRODE statement is used with the X and Y parameters acting as crosshairs to target a particular region of the structure. The whole region irrespective of shape is then defined as an electrode.

ELECTRODE NAME=gate X={x value} Y={y value}

A region defined this way is now treated as equipotential in ATLAS. The potential of this region will be defined by the VGATE parameter of the SOLVE statement. Hence poly depletion cannot be modeled in gate contacts defined this way. The workfunction of this region must be set by the user. For example a heavily n doped polysilicon contact can be defined by either of the two following statements:


The second approach of treating the polysilicon gate region as a semiconductor is achieved by placing a contact on the top of the gate. In ATHENA this is done be depositing a metal (or silicide) layer on top of the polysilicon. The ELECTRODE statement is then used to define this metal region as the gate electrode. In ATLAS a workfunction for the gate should not be specified on the CONTACT statement as this would give an undesirable workfunction difference between the metal and polysilicon. the potential on the metal region is defined by the VGATE parameter of the SOLVE statement. The potential within the polysilicon gate region will depend on the doping level of the polysilicon. In non-degenerately doped polysilicon a voltage drop is seen across this region from top to bottom. The workfunction difference between the gate and the substrate can be derived from a potential profile through the channel region.

It is also possible for the polysilicon to be depleted starting at the gate oxide interface. Figure 1 shows a comparison of high frequency CV curves between a MOS device with a uniform degenerately doped poly gate typical when tube doping is used and a lighter, non-uniformly doped gate typical when source/drain implants are used to dope the polysilicon. In the accumulation region the poly begins to deplete leading to an effectively thicker gate dielectric. This effect is illustrated in Figure 2. The amount of poly depletion observed is dependent on the doping level. Accurate polysilicon diffusion models are available in ATHENA to simulate the doping.1 In addition the SILICIDE module allows simulation of the dopant redistribution during gate silicidation. Silicides can typically reduce the effective gate doping making poly depletion more likely.


Figure 1 High frequency CV curve showing poly depletion effects at positive Vgs. Figure 2 Electron concentration profile of an NMOS transistor poly depletion occurs at the poly/gate oxide interface.