Silizium (Si-MOSFETs)
In collaboration with the group of
Prof. K. Hess and
Prof. U. Ravaioli
at the
Beckman Institut of the
University of Illinois
(
National Center for Computational Electronics) the EAs were used to
investigate the distribution of hot electrons in silicon.
The object was to calculate sustrate and gate currents in Si-MOSFETs.
With help of a physical mutation operator, which is based on the Monte-Carlo
technique, results comparable to full band calculations could be
obtained.
-Calculation of hot electron distributions in silicon by means of
an Evolutionary Algorithm,
J. Jakumeit, U.~Ravaioli, K.~Hess, J. Appl. Phys., Okt. 96,
(postscript, gezipt, 153kb)
-Evolutionary algorithms for the calculation
of electron distributions in Si-MOSFETs, J, Jakumeit, Proceedings of the
IV. International Conference on
Parallel Problem Solving from Nature (Berlin 1996), Lecture Notes in Computer
Science 1141, Springer Verlag, 1996, S. 819
(postscript, gezipt, 186kb)
Mutation-Operator-Monte-Carlo method
(MOMC)
An important part of the quasi backward calculation of electron
distributions by EAs is the physic based mutation.
Calling the physical mutation operator without using the optimization
of the EA leads to a new type of Monte-Carlo technique, the
Mutation-Operator-Monte-Carlo method (MOMC).
For the calculation of the distribution of hot electrons in bulk silicon
a better resolution of the high energy tail of the distribution could be obtained
by the new method when compared to a
Full-Band-Monte-Carlo-simulation. The MOMC requires also less time to
compute.
-Simulation of
Si-MOSFETs with the Mutation Operator Monte Carlo Method,J. Jakumeit, A. Duncan, U. Ravaioli, K. Hess,
Proceedings of the 5th International Workshop on Computational
Electronics (Notre Dame 1997), to be published in VLSI-Design
Local-Iterative-Monte-Caro-Technik
(LIMO)
The combination of many short and therefore local Monte Carlo steps
with and iteration process leads to the local iterative Monte
Carlo technique (LIMO). This new approach results in a isotropic
distribution of the computation time over the phase space, so that
regions with low carrier density are simulated with the same
accuracy as those regions with high density. The computation time
can significantly be reduced by memorizing the results of many
local Monte Carlo steps in a drift table. The LIMO technique can
easily be combined with an evolutionary optimization
algorithm.
The LIMO technique with and without optimization is base of the
ELIMO-package for the simulation of silicon and Si-MOSFETs. The
package includes all necessary informations and the source codes for
a test of this new Monte Carlo technique as well as some examples.
-Iterative Local Monte Carlo Technique for the Simulation of
Si-MOSFETs ,J. Jakumeit, T. Sontowski,
U. Ravaioli, Proceedings of the 6th International Workshop
on Computational Electronics (Osaka 1998), to be published in
VLSI-Design (postscript,
gezipt, 63 kb)
- ELIMO-package (getart,
gezipt, 2.7 Mb)