 |
 |
 |
Simulation of a 0.18 µm SI SOI n-MOSFET
 |
 |
|
Silicon-on-insulator (SOI) technology is under active study
as a candidate for future device technology.
Some images obtained from a study of hot electron effects in
a 0.18 µm SOI nMOSFET are shown below, to highlight the
ability of DAMOCLES to model potentially
important, "realistic" device
structures.
SOI nMOSFETs differ from the bulk counterparts (e.g., the preceding
nMOSFET and pMOSFET examples)
by the lack of a substrate contact. Instead, the active layer sits on
a "floating" oxide substrate.
The image below shows the electron and hole particle densities
in the SOI device cross section.
960×768 jpeg
The device bias is Vgate=2 and Vdrain=0.8 V, with the source grounded.
The temperature is 300 K.
Electrons and holes are colored according to their kinetic energies.
Electrons range from cyan (0 eV) through green and yellow, up
to red (1.5 eV). Holes span a much smaller energy range, and
are primarily magenta in color (0 to 0.3 eV, labeled -0.3 eV in the
figure as hole energy is negative by convention).
As electrons flow from source to drain, they impact ionize
near the drain end of the channel. The created electrons
flow out of the drain, but the created holes remain in the substrate.
There is no substrate contact, and hence, no direct path to permit the
excess holes from leaving the device. Instead, the holes pool near the
source. This alters the substrate potential, and can have important
effects on SOI device behavior.
Note how the (poly-)silicon gate has been modeled
by electron particles in (single crystal) silicon, in order to
allow the possibility of observing gate depletion effects.
The conduction band potential corresponding to this SOI device
and operating condition is shown below.
960×768 jpeg
The semiconductor potential is shown in shades
of blue, while the oxide regions jump to shades of yellow and red
due to the 3.1 eV discontinuity between the potential in the two regions.
Here is a plot of the total energy in the conduction band,
showing the position of electrons in the Si active layer of the
SOI device.
960×768 jpeg
The Si-substrate oxide interface corresponds to the front edge
of this figure, while the rear edge corresponds to the
Si-gate oxide interface.
Electrons move from the source (left) to the drain (right), and are
colored according to their kinetic energy.
There is a pillar of hot carriers near the drain: this is an artifact
of the manner by which DAMOCLES exaggerates rare electron states
in order to probe highly improbable portions of the distribution
function (like the high energy tail near the drain, in this case).
The plot below shows a similar image, except holes in the valence band
are plotted.
960×768 jpeg
The valence band axis is inverted, so that hole energy
increases from the bottom to the top of this plot. The holes are
quite low in energy, and congregate in the potential
energy well in the active layer near the source junction.
Quantitative carrier distribution information is readily available
from DAMOCLES simulations. Below is plotted the kinetic energy
distributions of electrons near the drain in this SOI device, for
two different modeling conditions.
851×801 jpeg
Of note of how different the high energy "tails" of the two distributions
functions are. Above 1.25 eV, electron-electron scattering is seen
to add many more energetic electrons, compared to the modeling
result without electron-electron scattering.
Physically, as electrons traverse the channel and then enter the
drain, electron-electron scattering provides a mechanism for
carriers to gain energy, as well as lose it. While the majority will
lose energy, a few (maybe 1 out of 1 million, or so) will gain
energy. Quantifying this tiny population of hot electrons is important in
estimating device reliability.
It may seem counterintuitive that electrons with an energy of 1.5 eV and more
exist in a device with only 0.8 V drain bias...but its true!
The page discussing low-bias effects provides
more details on the physical origin of these unexpected hot electrons.
damoclesNO-SPAM@watson.ibm.com
(last updated: January 26, 1999)
|