next up previous contents
Next: C. Analytical MOS Model Up: Michael Stockinger's Dissertation Previous: A. Drive Current as


B. Device Generator Makedevice

The program Makedevice is a tool for creating device descriptions of MOS devices with analytical doping profiles that can be used for device simulation. The supported MOS device structure is a typical bulk device with a planar gate and two sidewall spacers for source and drain (see Fig. 3.2).

The tool Makedevice first reads a template file in ASCII PIF format with the geometry points, grid points, and doping values being substituted by keywords, and completes the PIF file for these lacking values. Thus, the device structure is determined by the template file whereas geometry, grid, and doping information are computed by Makedevice. An input deck provides all necessary keywords. The hereby generated device description can directly be used as an input for MINIMOS-NT.

Several geometry parameters can be defined like the gate length, the gate oxide thickness, the lengths of the source and drain wells, and the thicknesses of the bulk and contacts.

There are three methods to define the acceptor and donor dopings which can be used coincidently:

Any number of Peak or Delta Dopings can be used together, all doping contributions are superposed. The Knot Doping Method defines the background doping inside a specified rectangle at the semiconductor surface. Outside this rectangle the background doping is given by a constant substrate doping.

For the Knot Doping Method a given number of knots is distributed homogeneously, just like an orthogonal grid, over a rectangular area. The boundaries of this area can be set by the user, except for the upper boundary which has to meet the bulk surface. The number of sections in $x$ and $y$ direction can be set arbitrarily. This method allows for generation of a user defined two-dimensional doping structure suitable for a very general optimization approach.

The doping is interpolated using two-dimensional orthogonal splines or the raised-cosine method which uses two-dimensional weight-functions having the shapes of raised-cosine pulses and widths of two knot sections. For each knot the doping value is multiplied by this weight-function and all contributions are added together to receive the final doping (see also Section 3.1.4). The doping values at each knot are given by a doping matrix in the input deck.

The Peak Doping Method can be used for generating doping peaks with two-dimensional joined-half-Gaussian or joined-half-raised-cosine shapes. An arbitrary number of Peak Dopings can be used (see also Section 3.1.5). The Delta Doping Method can be used for generating one-dimensional vertical doping profiles. It works similar to the Peak Doping Method, but without any lateral dependence of the doping.

The generated grids are ortho-grids. There are two grids generated by Makedevice: The doping grid and the simulation grid whose densities are controlled by a number of input deck parameters.

The density of doping-grid lines is chosen to represent the doping with a sufficient resolution. It can optionally be substituted by the simulation grid which is highly recommended considering that MINIMOS-NT interpolates the doping from the doping grid onto the simulation grid during the read-in state. But a denser doping grid can, for example, be used for visualization purposes.

The simulation grid must fulfill two criteria simultaneously: It must represent the doping with a sufficient resolution for simulation purposes and its density must be high in junction regions and in regions where the resulting potential and carrier-density drops are high. The latter is true for the channel region and the source-channel and drain-channel junctions.


next up previous contents
Next: C. Analytical MOS Model Up: Michael Stockinger's Dissertation Previous: A. Drive Current as
Michael Stockinger
2000-01-05