Information Technology Reference
In-Depth Information
Efforts have been made in recent years on modeling CNFETs [63, 64]
and CNT interconnects [65, 66] to evaluate the potential performance at the
device level. Very promising single device DC performance over silicon devices
has been demonstrated either by modeling or experimental data. However, the
dynamic performance of a complete circuit system, consisting of more than one
CNFET and interconnects, differs from that of a single device. All but one of the
reported models to date used a single lumped gate capacitance and ideal ballistic
model to evaluate the dynamic performance, which results in an inaccurate
prediction [67, 68]. To evaluate CNFET circuit performance with improved
accuracy, a CNFET device model with a more complete circuit-compatible
structure and including the typical device nonidealities was constructed [10].
This recent publication presents a novel circuit-compatible compact SPICE
model for short channel length (5 l nm 100 l nm), quasi-ballistic single wall
carbon nanotube field-effect transistors (CNFETs). This model includes practical
device nonidealities, e.g., the quantum confinement effects in both circumfer-
ential and channel length direction, the acoustical/optical phonon scattering in
channel region and the resistive source/drain, as well as the real time dynamic
response with a transcapacitance array. This model is valid for CNFETs for a
wide diameter range and various chiralities as long as the carbon nanotube
(CNT) is semiconducting.
17.3.3. Measurement and Modeling of Cortical Data
Significant cortical data has been collected by Braitenberg and mathematically
categorized [69, 70]. Recent research by Stevens at the Salk Institute [71] relates
the volume of an axonal arbor to the total length of the axon branches in the
arbor. This type of information can be useful in predicting interconnection
requirements for a synthetic cortex. Sejnowski [72, 73], also at the Salk Institute,
has focused on statistics concerning neural communication.
3D biological data obtained in collaboration with Manbir Singh (USC
Department of Biomedical Engineering) provides statistical information for our
prediction models [74]. In the nanoelectronic approach case, Singh's three-
dimensional statistics can be applied directly to provide insight into the inter-
connectivity problem for the synthetic cortex. We postulate that the distance
presynaptic terminals are from the soma can be modeled as an exponential
(Poisson) distribution, with the axon more likely to have presynaptic terminals
close to the soma. Experimental results [74] support this assumption. We also
postulate, similar to Steven's work at Salk, that there is a Rent's rule for regions of
the cortex, relating statistically the number of axons emerging from a spherical
volume of brain tissue to the volume of brain tissue enclosed.
Bailey and Hammerstrom provided early predictions on interconnection
possibilities using CMOS technology [75],
including an early reference to
Sejnowski's research.
 
Search WWH ::




Custom Search