SPRINTCAD QUARTERLY SUMMARY
Oct. 1 - Dec. 31, 1996
ORGANIZATION:
Stanford University
SUBCONTRACTORS:
none
PRINCIPAL INVESTIGATORS:
Robert W. Dutton, dutton@gloworm.Stanford.EDU, (415) 723-4138
Kincho H. Law, law@cive.Stanford.EDU, (415) 725-3154
Krishna Saraswat, saraswat@sierra.Stanford.EDU, (415) 725-3610
Peter Pinsky, pinsky@cive.Stanford.EDU (415) 723-9327
TITLE OF EFFORT:
"SPRINT-CAD"---Industry-Networked TCAD using Shared Parallel Computers
RELATED INFORMATION:
The URL for Stanford TCAD projects is: http://www-tcad.stanford.edu/
The URL for the Sprint-CAD project is: http://www-tcad.stanford.edu/sprintcad/
OBJECTIVE:
First-time capabilities to bridge solid modeling, FEM-based
parallel computation of fabrication processes and electrical analysis
of the resulting IC structures will be developed. Models needed
to represent diffusion, etching, deposition, oxidation and stress
analysis resulting from a sequence of process steps necessary in
the creation of electrical devices will be developed. This effort
will provide a radically new HPC framework for technology-based 3D
process/device modeling as well as realistic benchmarks to test HPC
architectures and software.
APPROACH:
We will build, integrate and test TCAD modules based on
an object-oriented approach that both develops and uses information
models in support of CFI-based standards. The modules and software
engineering methodology will be designed specifically to exploit
parallel computers and library components. The 3D process simulation
modules will utilize HPC platforms and provide new functional
capabilities for "computational prototyping" of the following key
technology fabrication steps: 1) deposition/etching module---of special
interest are CVD and plasma assisted processes that result in high
aspect ratio structures such as trenches and filling/planarization
of structures for metal interconnects. Algorithmic work focuses on
geometric manipulations and surface evolution. 2) thermal/stress
analysis module---that can solve nonlinear constitutive models for
key process steps involving growth of dielectric layers and impurity
redistribution as well as the resulting stress fields. Advanced
formulations for finite elements are being developed that support:
parallel computation, adaptive gridding and domain decomposition.
PROGRESS:
It is established that the waveform relaxation technique is
stable and efficient for large set of ordinary differential
equation (ODE) systems as in circuit simulation. Parallelization
based on waveform relaxation is attractive since the calculation
on subsystems can be asynchronous. Application of waveform
relaxation for partial differential equations (PDE) with subsystems
partitioned by domain decomposition with controllable overlap
regions is very promising for parallelization of the reactive-diffusive
systems, since stability and rate of convergence are directly
related to the maximum principle (the ordering principle if the reactive
term is strong).
Fast prototyping of new physical models using ALAMODE has been
demonstrated for boron diffusion. The new boron model includes
reactions of point defects and defect-dopant pairs, considering their
charge states, and the dopant inactivation by the introduction of a boron
clustering reaction. The six transport equations and complex
constitutive models for diffusive and reactive terms have been completely
described in the TCL extension language without any modification and
compilation of the source code. Rate of convergence and level of
accuracy are comparable to the original implementation for the specific
model.
Simulation results for oxidation have been demonstrated with
finite deformation kinematics and level-set boundary movement.
Special incompatible element functions have been used to represent
the discontinuity of transport coefficients at the material interface.
This algorithm has eliminated the constraint of gridding conformity
to the interface, which has caused excessive gridding and sometime
failure of convergence in 2D and 3D. The curvature can be accurately
calculated and hence the stress analyses for complex structures can
be more stable and accurate.
The minimal semiconductor representation (SWR) tries to capture the
sufficient set of functions (methods) for TCAD tool communication
and functionality sharing. With new inclusion of established software
such as EUCLID (unstructured triangular/tetrahedral gridding) in the
SUPREM OO7 architecture, applicational procedural interface (API)
for geometry/field servers has been slightly modified to efficiently
handle multiple volumes and fields. The ALAMODE/server interface
has also been finalized.
The level set method for boundary movement is stable and accurate
according to our investigation effort in physical model calibration
in the previous period. However, since volume grid instead of surface
grid is used, computational efficiency has been a concern even if
adaptivity is used. Efficiency benchmark on etching/deposition
simulation has been established for various physical models
and staggering schemes. The most CPU intensive part is at the
transformation between boundary representation, which is required
only if re-emission or surface diffusion is included. However,
the solution of the large linear system of equations dominates the
entire simulation when those models are included, and hence the
computational penalty using the level set method is not significant.
RECENT ACCOMPLISHMENTS:
- Waveform relaxation algorithms for partial differential equations
(PDE) have been applied to the study of domain decomposition
of reactive-diffusive systems.
- ALAMODE has been applied for Boron nonequilibrium diffusion
model, which accounts for reactions of point defects and
defect-dopant pairs. (Kunikuyo and Dan)
- Simulation results for oxidation have been demonstrated with
finite deformation kinematics and level-set boundary movement.
Special incompatible element functions have been used to represent
the discontinuity at the material interface.
- Application procedural interface to EUCLID (unstructured tetrahedra
gridding) for multiple volumes and fields has been implemented
according to the minimal semiconductor wafer representation
(SWR).
- Efficiency benchmark for level set methods on etching/deposition
simulation has been established for various physical models
and staggering schemes.
FY-'97 PLANS:
- Parallelization of ALAMODE and the geometry/field server functions
through the static domain decomposition techniques.
- 3D conformal gridding tools to provide 3D elements that satisfy the maximum
principle in various mass lumping schemes for the diffusive-reactive systems.
- Validation of the 1-2-3D geometry/field server using fully adaptive schemes
for semi-discrete finite-element based TCAD tools.
- Benchmark the overall SPRINT-CAD modules in terms of a deep
submicron MOS technology. The thermal module will demonstrate 3D
analysis of locally oxidized isolation and shallow junction diffusion
steps. The deposition/etching module will demonstrate first-time
functionality of a 3D server-based architecture and implementation.
TECHNOLOGY TRANSITION:
The test sites of ALAMODE have been continuously growing, with additions
of the groups at Texas Instruments, Boston University and MIT.
More positive responses on the model flexibility have been fed back
to the developers at Stanford. A short course on ALAMODE is planned
in the near future.
Gridding has perpetually been a difficult part for practical TCAD
applications. Requirements on gridding for accurate 2D/3D simulation
have been discussed with Intel, primarily on the MOS transistors
and isolation structures. Gridding tools developed at Stanford such as
CAMINO and EUCLID, together with the SWR API interface, have been
considered for resolving the complex geometry features.
Edwin C. Kan
kan@gloworm.stanford.edu
CIS-X 334, Stanford University, Stanford, CA 94305
Office: (415) 723-9796
Fax: (415) 725-7731
Date prepared: 10/31/96