The device is based upon the super-halo discussion by Taur et al.  and estimated structural details from the Road Map. Reasonable device geometries can be designed for simulations in the deep sub-100-nm regime. The gate has the n+ workfunction and the effective tox is 9.3 Angstroms (which is equivalent to the physical tox since poly depletion is not included here). Leff (as defined between where the source/drain dopings fall to 2e19 cm-3) was extracted and found to be 7 nm. Because the device is symmetric, the origin for the lateral grid is at the middle of the channel (x = 0); the interface between the gate oxide and bulk is chosen as the depthwise origin (y = 0). The source and drain contacts have an added external lumped resistance that gives a total Rsd of 80 Ohm.um to the device.
A 2-D doping profile was developed. There is a separate background uniform p-type doping of 1e17 cm-3. The S/D and halo profile is simply generated from an analytic formula, as in sh9.analytic, that describes two 2-D gaussian functions: one for the n+ source/drain, one for the p+ halo. The device profile is mirror symmetric about x = 0. The lateral slope of the S/D extension falls off at 1 nm/decade while the halo falls off at 2 nm/decade as seen in a lateral cut of the doping profile along the surface: Lateral
A set of simulated Id vs. Vgs curves (PREVIEW) are stored in the idvg
directory. The characteristics have been generated using the Energy-Balance
(EB) model as described in MEDICI . To speed convergence, no quantum mechanical (QM) correction or poly depletion has been used; however, we expect altered I-V's once the "electrostatic" QM effects are included. Properly including those as well as QM effects in transport (Gate-Channel leakage, S/D tunnelling, and Band-to-Band tunnelling) is left to the user as an exercise. A sample MEDICI file
inputfile9 is included for reference. The I-V characteristics are text files
labeled according to their drain bias (where Vbs = 0 V); for example, d0.1
has a Vds of 0.1 V. Each file has two columns of data: the first is Vgs in
0.1 V steps and the second is normalized Id (A/µm). The Ioff of the simulated device is lower than the Road Map target of 10 uA/um at an operating bias of 0.4 V. However, the DIBL seems constrained to about 300 mV/V because there is a limit to how much the p+ halo can ameliorate the drain induced band bending.
 Y. Taur et al., IEDM proceedings, p. 215, 1997
 MEDICI manual, Technology Modeling Associates