 |
 |
| . |
. |
Unique Sony Design Technologies that Support Image Sensor Development
Image Sensor Simulation Technology |
|
3D simulation |
|
Simulation accuracy |
|
GUI* that increases simulation efficiency |
|
Adopting the design of
experiments method
for increased efficiency |
| |
* GUI: Graphical User Interface |
Image sensors make the number of applications
such as camcorders, digital still
cameras, and cellular phone, that includes
itself greater. Pixel is scaled down year
by year according to high resolution that
make high-magnification digital zoom
possible. Smaller pixel sizes, however,
make it increasingly difficult to provide
good device characteristics, such as
sensitivity and saturation signal, and furthermore
affect the other characteristics
that were previously not thought to be
influenced. Another issue here is that as
the end product development cycle becomes
shorter, the device development
period is also becoming shorter every
year. Thus given these conditions, the
semiconductor simulator, which makes it
possible to shorten development periods
and reduce development costs, has become
indispensable in contemporary
semiconductor device development.
When Sony first began to develop image
sensors, they developed independently
their own dedicated image sensor process
and device simulators in an effort to predict
device characteristics and shorten the
development period. In this article, we
present an overview of Sony’s process and
device simulation technologies that support
Sony’s CCD development. |
|
|
 |
3D Simulation |
| . |
. |
The part of the image sensor structure that
is its “eye”, i.e. the pixel, has a structure
that differs from that of a MOS transistor
in that it has a pattern that is asymmetric
in both the left/right and up/down direction.
As a result, the 2D simulation techniques
used for MOS transistors cannot
be used for image sensors. Therefore
Sony, very early on, developed their own
3D simulator and have been using it in
image sensor development ever since.
This 3D simulator was developed to
match the features of the CCD process and
device structure, and, from the
process aspect, adopts a physical model
that takes into consideration diffusion
during high-temperature oxidation and
aims at speeding up the calculations
concerning low-concentration ion implantation.
From the device standpoint, this
simulator adopts algorithms that can calculate
the boundary conditions as periodic
conditions, since the CCD pixel pattern
is periodical.
To improve ease of use, this simulator was
implemented so that it can accept simulations
with multiple drive conditions (the
same structure but different bias conditions)
in a single input file. Using thus 3D
simulator to predict device characteristics
and determine prototyping conditions can
contribute to shortening the prototyping
period significantly and reducing development
costs.
|
|
|
Simulation Accuracy |
| . |
. |
After accumulating charges from the
photodiodes, a high voltage is applied to
the readout electrode and the accumulated
charges are read out to the vertical
register in the CCD. Since a high voltage
is applied to the electrode, a high insulation
film breakdown voltage is required.
Although the electrode insulation film
was previously formed by a high-temperature
oxidation process, it has become
necessary to suppress the thermal diffusion
of impurities in the silicon due to the
scaling down the devices. Associated with
this, Sony has switched to a simulator that
has models capable of highly accurate
prediction of the doping profile due to
thermal diffusion. When using this new
simulator, Sony has made a point of
making the doping profile calculated by
the simulator match the doping profile in
the actual devices, and thus achieving
even higher simulation accuracy. |
|
|
Increasing Simulation
Efficiency |
| |
|
No matter how good the accuracy of the
simulator, if it isn’t easy to use, it won’t
be able to withstand the demands of practical
activities in the design process. Sony
made a variety of improvements in operability
and speed in the simulation design
process flow. First, note that the simulation
operation can be divided into the following
three operations.
1. Creating the simulation input files
2. Running the calculation
3. Analyzing the results
Since the simulation input data is 3D, it
is complex and difficult to create, and just
coding this data is a time-intensive
operation. Since it is a manual operation,
coding mistakes can occur easily. To
resolve these issues, Sony went to the
effort of developing a GUI environment
that creates the simulation data from mask
patterns from the CAD layout tools and
process condition tables. In this GUI
environment, the designer creates the
simulator input files by selecting the mask
pattern and process conditions and setting up the simulation conditions. Thus the
designer can perform all required operations
through running the simulation even
if the designer does not know the coding
method used to create the simulation files.
This makes it possible to perform in a few
hours the work that previously required
several weeks of hand labor. The simulator
itself is implemented so that simulations
are executed using optimized
algorithms that take advantage of the
features of the latest hardware and parallel
processing. This results in simulation
speeds 20 times faster than earlier systems.
Although earlier simulators had
limitations on the number of process steps
due to simulator hardware restrictions, the
upper limits have been improved greatly
so that they are appropriate for today’s
hardware. Sony also developed a GUI so
that the high volume of critical data produced
by the simulator can be analyzed
appropriately. Sony is developing a GUI
for efficient statistical processing of data
acquired from making a variety of
changes to the process and device conditions.
By graphing the changes in the
device characteristics as the conditions
change, designers will be able to immediately
grasp the device characteristics. |
|
|
Simulation Design Using the
Design of Experiments Method |
| . |
. |
When using a simulator to determine
prototype conditions, Sony simulators
now perform simulations based on the
DOE (Design of Experiment) improve
efficiency even further. For example, with
two levels and eight factors, it would be
necessary to perform 28 = 256 simulations.
However, device characteristics can
be predicted with just 16 simulations
using the DOE. It is possible to quickly
determine design conditions that will meet
the specifications for multiple device
characteristics by parallelizing the regression
equations acquired by the analysis
of a variety of characteristics. Also, the
yield can be forecast by adding process
variation to the parallelized equations.
Sony has developed tools that perform the
steps from analysis through determination
of optimal conditions, and has significantly
reduced the development time up
to prototype creation. |
|
|
Future Developments |
| . |
. |
The image sensor market is expected to
continue to grow even faster in the future.
Given these conditions, it is clear that the
process and device simulation technologies
presented here will become even
more important in the future. Sony is
aiming for even further improvements in
accuracy and designer work efficiency,
and is working on matching simulator results
to actual devices and constructing
even more efficient simulation environments.
You can look forward to even more
advanced image sensor process and
device simulation technologies from Sony
in the future. |
|
 |
See
all articles with figures and tables.  |
|
Vol.42 |
|
|
|
|
