MOS Modeling and Parameter Extraction Group
MOS-AK Meeting
organizers: Fabien Prégaldiny, Christophe Lallement and Wladek Grabinski
in cooperation with Institut d'Electronique du Solide et des Systèmes and Ansoft
8 April 2005, Strasbourg
|
|||||||||
|
|
|||||||||
| 9:25 - 9:30 | W.Grabinski Introduction |
||||||||
| 9:30-11:00 |
Morning
Session Chairmen: Fabien Prégaldiny and Wladek Grabinski |
||||||||
| Daniel
Mathiot (InESS): Welcome address and research activities at InESS |
|||||||||
| Patrick
Martin,
Mickaël Boasis,
Olivier
Rozeau, Jérôme Prouvee and David
Axelrad
(LETI/CEA-Grenoble): Wafer-Level Extraction of BSIMSOI Low Frequency Noise Parameters for 130 nm Partially-Depleted SOI MOSFETs |
|||||||||
| Luc Hebrard, J.-B.
Kammerer, M. Hehn, V. Frick, A. Schuhl, P. Alnot, P. French and F.
Braun (InESS): CMOS compatible integrated magnetometers |
|||||||||
| Jean-Claude
Perraud
(ENSI CAEN): New Verilog A compiler for SPICE 3F5 |
|||||||||
| Coffee Break | |||||||||
| 11:15-12:00 |
Poster
Session Chairmen: Jean-Michel Sallese and Wladek Grabinski |
||||||||
| D.
Jiménez, B.
Iñíguez, J. Roig, J. Suñé,
L. F. Marsal, J.
Pallarès and D. Floresc (U.Barcelona): Analytical continuous I-V model of the Surrounding-Gate MOSFET |
|||||||||
| Jean-Baptiste
Kammerer,
L. Hebrard and F. Braun (InESS) Implementation of an hysteresis model in VHDL-AMS for compact modeling of spintronics devices |
|||||||||
| Larry
Dangremond
(Cascade): Eliminating radio frequency interference effects on 1/f, Vt, gm and capacitance measurements |
|||||||||
| Alain
MICHEL (Ansoft
Europe): Nexxim®: Ansoft’s New State-of-the-Art Circuit Simulator for RF, Analog and Mixed-Signal Design |
|||||||||
| Fadhila Haned, M. Ben
Chouikha,
G.Alquié (LISIF, Pierre et Marie Curie University, Paris): An Improved BDJ Color Detector Physical Model |
|||||||||
| M. Alwan, B. Beydoun, K.
Ketata, M. Zoaeter
(Rouen University IUT-LEMI) 2D analysis of a power VDMOS transistor CV characteristics temperature effects |
|||||||||
| 12:00-13:30 |
Lunch |
||||||||
| 13:30-16:30 |
Afternoon
Session Chairmen: Christophe Lallement and Wladek Grabinski |
||||||||
| Franz Sischka (Agilent): EM substrate effects modeling using Agilent Momentum |
|||||||||
| Birahim
Diagne, Fabien
Prégaldiny, François Krummenacher,
François
Pécheux, Jean-Michel Sallese and Christophe Lallement (InESS/EPFL/LIP6) Design Oriented Model for Symmetrical DG MOSFET |
|||||||||
| Thomas
Zimmer (IXL
Bordeaux): Self-heating investigation of bulk and SOI transistors |
|||||||||
| D.
Rideau, F. Gilibert
and M. Minondo (STM): Modelling Strained Silicon Devices |
|||||||||
| Johannes
Fellner
(austriamicrosystems AG): A CMOS compatible PolyFuse element used in a One Time Programmable circuit |
|||||||||
| 16:30 |
End of the MOS-AK Workshop | ||||||||
| Publication
Partner: International Journal of Numerical Modelling: Electronic Networks, Devices and Fields Editor, Europe: Phil Mawby  |
|||||||||
| Abstracts | |||||||||
| Daniel Mathiot (InESS): Welcome address and research activities at InESS InESS
(Institut
d'Electronique du Solide et
des Systèmes : Institute for solid state electronics and
electronics of systems) is a new laboratory resulting from the
association of two former laboratories, PHASE (Physics and Applications
of Semiconductors) and LEPSI (Laboratory of Electronics and Physics of
Integrated Systems). The new laboratory is a "mixed research unit" of
the Louis Pasteur University (Strasbourg) and CNRS (French National
Research Center). The research activities of InESS span various topics,
extending from solid state electronics to electronics of systems. The
different themes are thus concerned with materials and nanotechnologies
for electronic devices, materials and concepts for photovoltaics,
optoelectronic instrumental systems and microsystems, and integrated
instrumental systems.
|
|||||||||
| Patrick Martin, Mickaël
Boasis,
Olivier
Rozeau, Jérôme ProuvÉe and David
Axelrad
(LETI/CEA-Grenoble): Wafer-Level Extraction of BSIMSOI Low Frequency Noise Parameters for 130 nm Partially-Depleted SOI MOSFETs Accurate
modeling
of low frequency noise
has become essential for downscaled analogue and RF circuits design,
because low frequency noise is upconverted to high frequencies through
the phase noise. Point-probe noise measurements enable the extraction
of BSIMSOI noise parameters (EF, NOIA, NOIB, NOIC and NOIF) referring
to physical µ- N flicker noise model based on
mobility correlation. Indeed, theoretical µ- N
model is implemented in BSIM code thanks to relationships linking the
oxide trap density Nt and the Coulomb scattering
coefficient to their equivalents in circuit
simulators. Nevertheless, measurements carried out on 130 nm PD SOI
MOSFETs suggest the need for a 103 factor in the equivalence
relationships according to the BSIMSOI version.
|
|||||||||
| Luc Hebrard, J.-B.
Kammerer, M. Hehn, V. Frick, A. Schuhl, P. Alnot, P. French and F.
Braun (InESS): CMOS compatible integrated magnetometers The
main physical
principles for magnetic
sensing leading to magnetometers which can be integrated into a silicon
die are reviewed . Then 1D, 2D and 3D integrated Hall effect magnetic
sensors in CMOS technologies are discussed with their potential
applications. These magnetometers are still integrated with a
surrounding electronics and the need for accurate compact models is
emphasized. In a second part, we present a 2D magnetic sensor based on
a Magnetic Tunnel Junction (MTJ). This new magnetometer is a good
candidate for integration into CMOS thru simple post-processing. This
sensor relies as much on the MTJ device as on its associated electronic
system. Here again, in order to simulate the whole magnetometer, the
need for a good compact model of the MTJ is emphasized.
|
|||||||||
| Jean-Claude Perraud
(ENSI CAEN): New Verilog A compiler for SPICE 3F5 From
a Verilog A
source model, it generates
all the C files that allows to manually install the new model into
Spice3F5,or using a new Spice command,load directly the new compiled
model into spice (using shared-lib mechanism). It has mainly all the
features of Verilog A needed for model generation, and some new
extensions to simplify the Verilog A sources: Verilog A Analog function
with InOut parameters, No return Value Analog function, you may define
separately device and model parameters, Macros may have more than 1
lines. It has been already validated for the following Spice analysis:
OP, DC, AC, TRAN, PZ, and related analysis. Noise analysis interface in
development. Also in development a new SPICE3F5 extension with RF
HARMONIQUE BALANCE,that includes S parameter Analysis,Port,Nport,HB
analysis(steady-state/1 forced source), QP analysis (steady-state/2
forced sources),CE analysis (TRAN HARMONIQUE BALANCE 2 forced sources).
Already, the New VERILOG A Model Compiler generates also the C files
for the RF HARMONIQUE BALANCE analysis.
|
|||||||||
| D. Jiménez, B.
Iñíguez, J. Roig, J. Suñé,
L. F. Marsal, J.
Pallarès and D. Floresc (U.Barcelona):
Analytical continuous I-V model of the Surrounding-Gate MOSFET A
continuous
analytic
current–voltage (I–V) model for cylindrical undoped
(lightly doped) surrounding gate (SGT) MOSFETs is described. It is
based on the exact solution of the Poisson’s equation, and
the current continuity equation without the charge-sheet approximation,
allowing the inversion charge distribution in the silicon film to be
adequately described. This model correctly traces the transition
between the different operation regions without resorting to fitting
parameters, being ideal for the kernel of SGT MOSFETs compact models.
We demonstrate that the I–V characteristics obtained by this
model agree with three-dimensional numerical simulations for all ranges
of gate and drain voltages.
|
|||||||||
| Jean-Baptiste Kammerer,
L. Hebrard and F. Braun
(InESS) Implementation of an hysteresis model in VHDL-AMS for compact modeling of spintronics devices Spintronics
devices such as magnetic tunnel
junctions or giant magnetoresistances have been recently integrated to
standard CMOS processes. This kind of devices can be used to design
random access memories, transistors or magnetic sensors. Nevertheless,
the electrical properties of spintronics devices depend on the
magnetization of their ferromagnetic layers and accurate hysteresis
models are needed for compact modeling of such devices. Here, such an
hysteresis model written in VHDL-AMS is described and a simple magnetic
tunnel junction compact model based on this hysteresis model is
presented. Simulation results of a MRAM cell and of magnetic sensors
are also shown.
Eliminating radio frequency interference effects on 1/f, Vt, gm and
capacitance measurements Some
measured
parameters are highly
susceptible to spurious electrical noise that impinges on a wafer
device. Example affected measurements are Vth & Id, gm, several
capacitance parameters, and of course flicker noise. The offenders may
be radiated or conducted emissions from external sources or locally
generated emissions. Improvements in isolation, shielding, power
supplies, and filters of on-wafer systems have been recently initiated
that enhance measurement accuracy , ease and speed . This presentation
shows examples of affected data and improved results.
|
|||||||||
| Alain MICHEL (Technical
Director, Ansoft
Europe): Nexxim®: Ansoft’s New State-of-the-Art Circuit Simulator for RF, Analog and Mixed-Signal Design Nexxim®
is
Ansoft’s new
circuit simulator that provides fast and accurate solutions for tough
analog and mixed-signal circuits. It
provides the transistor-level simulation accuracy required for
sensitive analog and wireless front-end circuits and the capacity to
solve the complexities in modern mixed-signal integrated circuits.
Transient and harmonic balance simulations can be performed from a
single netlist using the same
time-domain device models for broad simulation and analysis of analog
and RF circuits. New numerical algorithms and advanced
software engineering combine to deliver orders of magnitude improvement
in simulation speed, accuracy, and capacity compared to
other commercial simulators. Full compatibility to
HSPICE® netlist format provides full foundry device model
support
and
allows Nexxim to integrate into most EDA flows. Dynamic links
to the HFSS® electromagnetic field solver provides
on-chip passive and IC package modeling. Integration within
the Ansoft Designer® environment
provides schematic capture, netlist generation and editing, simulation,
dynamic links to system and planar EM
simulation, and results post processing. This presentation
introduces Nexxim’s capabilities and advantages and describes
its use within the general design flow of RF/Mixed-signal IC design.
|
|||||||||
| Fadhila Haned, M. Ben
Chouikha,
G.Alquié (LISIF, Pierre et Marie Curie University, Paris): An Improved BDJ Color Detector Physical Model The
BDJ (Buried
Double pn Junction) color
detector structure in a standard n-well CMOS technology consists of two
buried pn junctions. By using the optical properties of silicon, each
active area gives the color information without use of optical filters
The BDJ photocurrents spectral responses present two peaks in the
wavelength range 400 to 800 nm and the photocurrents ratio shows a
monotonically increasing function with the wavelength. Thus, both
wavelength and optical power of a monochromatic light can be
determined. This allows the use of the BDJ device for accurate
multispectral cameras and in imaging systems development. However,
successful design of color sensor using this detector requires an
accurate knowledge of it behavior. The physical model of the BDJ
photocurrents has been proposed in previous works. Even if simulation
results seem to give good fit to experimental data, the difference
between measurements and simulations remains important. In order to
develop a reliable model that simulates accurately the BDJ behavior, we
adopted a new modeling approach. The approach consists on experimental
extraction of most physical and electrical parameters. It takes also
into account the physical phenomena that intervene to form the BDJ dark
and photo-generated currents. Simulation results of both dark and
photo-generate currents are compared to measurements curried out on a
test chip designed and manufactured using a standard CMOS process. To
evaluate our model accuracy, we calculated at every wavelength the
relative error between simulation results and measurements. We note
that by using our modeling approach the difference is reduced. It is
about 1% in the wavelength range 490 to 750 nm.
|
|||||||||
| M. Alwan, B. Beydoun, K.
Ketata, M. Zoaeter
(Rouen University IUT-LEMI) 2D analysis of a power VDMOS transistor CV characteristics temperature effects TBD
|
|||||||||
| Franz Sischka (Agilent): EM substrate effects modeling using Agilent Momentum |
|||||||||
| Birahim Diagne, Fabien
Prégaldiny, François Krummenacher,
François
Pécheux, Jean-Michel Sallese and Christophe Lallement
(InESS/EPFL/LIP6) Design Oriented Model for Symmetrical DG MOSFET A
new
charge-based compact model for
undoped DG MOSFET under symmetrical operation is presented. This design
oriented model gives insight into the physical phenomena and is
especially dedicated to the circuit designer. Based on the EKV
formalism, useful normalizations are proposed and successfully
implemented in the model. This allows to greatly simplify the model
formulation in deriving current and charges in a very comprehensive
form. In order to validate the analytical model, we have also developed
the 2D simulations of a DG MOSFET structure (tox=2nm, tsi=25nm) and
performed both static and dynamic electrical simulations of the device.
Comparisons with the 2D numerical simulations give evidence for the
good behavior and the accuracy of the model. In particular, the ideal
sub-threshold slope (60mV/decade) that takes place in the
weak-inversion region (also called the "volume inversion region" for
the DG MOSFET) is well described. Finally, the VHDL-AMS implementation
of the DG MOSFET model is carried out leading to fast and efficient
simulations.
|
|||||||||
| Thomas Zimmer (IXL
Bordeaux): Self-heating investigation of bulk and SOI transistors This
presentation
deals with the
self-heating effect in bulk and SOI transistors. The static and dynamic
self-heating mechanism is investigated. The resulting theoretical model
is implemented in an electrical equivalent circuit. The approach is
validated through measurements on devices from different technologies.
System configuration, measurement, and calibration issues are
presented.
|
|||||||||
| D. Rideau, F. Gilibert
and M. Minondo (STM): Modelling Strained Silicon Devices Modelling
the
mechanical stress-induced
effects on MOSFETs capacitance and currents becomes a crucial issue for
device simulation. From the one hand, the scaling down of MOSFETs gives
rise to parasitic mechanical stress such as sidewell oxidation volume
expansioninduced stress. One the other hand, strained Si is becoming an
essential component in device application due to its enhanced carrier
mobility. Strained Si on relaxed SiGe buffer layer or film-induced
stress are typical techniques used to improve device performance. Based
on numerical simulations of the electronic band structure, we review
the impact of the stress on the first order MOSFETs parameters. We also
show that some basic parameters such as the effective mass or the band
gap are strongly stress-dependant.
|
|||||||||
| Johannes Fellner
(austriamicrosystems AG): A CMOS compatible PolyFuse element used in a One Time Programmable circuit A
Poly Fuse
element, which is based on a
2-layer approach, was developed to be used as an 'One Time
Programmable' element in a 0.35um standard CMOS process. Optimal
programming conditions are defined as well as drifts of the programming
conditions and their impact on the Poly fuse reliability are monitored.
TEM/SEM analysis gives corresponding physical explanations for the
electrical effects seen for different programming conditions. For
process control the Poly Fuse and the corresponding programming
transistor is implemented into the standard Scribe Line Monitor. So it
is possible to get the variation of unprogammed and programmed values
for statistics. The implementation of the Poly Fuse into a circuit
takes care of all reliability restrictions and is focussed on a high
programming yield. Due to the requirement of larger numbers of bits,
the OTP structure is combined with a RAM structure to share the control
logic. A state machine handles the interaction and has implemented also
a startup sequence. After startup, the block has a RAM compatible
access to the data. Some defined bits can be used directly for trimming
purpose.
|
|||||||||
 |
|
|