Generated compact models with improved accuracy - MOS-AK

Download Report

Transcript Generated compact models with improved accuracy - MOS-AK

MOS-AK Meeting, September 22, 2006
Application of physicsbased device models for
circuit simulation
Victor Spitsyn, Ilya Lisichkin
Cadence Design Systems LLC, Moscow
A Model: equations and parameters
The model derivation includes
discretization mesh, Poison and
continuity equations integration
steps. A circuit representation is
derived. Some simplifying
assumptions were done.
δQkp
Ckp = ---------------δ(Vkp – Vki)
Jkp = e gkmn (Vk-1i – Vki)
* gkdn (Vk-1i – Vk-1n, Vki – Vkn )
* gkvn (Vk-1n – Vkn)
Pn junction circuit model
Three-block equivalent circuit of
a pn junction. The elements
drawn in black are those which
implement the minimum device
functionality. The elements
drawn in grey are only required
for accurate high-injection and
high-frequency modeling. The
heavy dark lines highlight
elements which reduce to short
circuits in the low-injection case.
Simulation of pn junction
Small-signal conductance and
capacitance at 100kHz for a
symmetric pn junction with 5μm
long p and n regions of doping
1016cm-3. The order parameter
indicates the number of partitions
on each side of the junction.
• Frequency response (real and
imaginary part of the small-signal
admittance) for the first- and
second-order model, at a forward
bias of 0.7V.
Npn bipolar transistor circuit model
Equivalent circuit of the
npn bipolar transistor. The
elements drawn in black
are those which
implement the minimum
device functionality. The
elements drawn in grey
are only required for
accurate high-injection
and high-frequency
modeling. The heavy dark
lines highlight elements
which reduce to short
circuits at low injection
Simulation of bipolar transistor
Frequency dependence of the real and imaginary parts (absolute
values) of the Y parameters for a 0.25 μm npn bipolar transistor
at a current density of 430 μA/ μm2 (ft = 16 GHz, fmax = 30 GHz)
What’s next: performance and
accuracy gain measurement
• Using common benchmarks and typical engineering tasks to
measure performance and compare accuracy wrt compact models
Summary
• Considered models are good to apply for small-signal
analyses.
• For successful implementation, parameter extraction tool
is needed, because optimization procedure is an
essential part of the entire flow.
• Assessment of performance impact and accuracy gain
as compared to the compact models is reasonable to
complete.
References
• J. G. Linvill, “Lumped models of transistors and diodes”,
Proc. IRE, vol. 46, pp. 1141–1152, June 1958.
• C. T. Sah, “The equivalent circuit model in solid-state
electronics—Part I: The single energy level defect
centers”, Proc. IEEE, vol. 55, pp. 654–671, May 1967
• A. Pacelli, M. Mastrapasqua, and S. Luryi, “Generation
of equivalent circuits from physics-based device
simulation”, IEEE Trans. Computer Aided Design of
Integrated Circuits, vol. 19, pp. 1241–1250, Nov. 2000.