SPRINTCAD QUARTERLY SUMMARY
October 1 - December 31, 1995
ORGANIZATION:
Stanford University
SUB-CONTRACTORS:
none
PRINCIPAL INVESTIGATORS:
Robert W. Dutton, dutton@gloworm.Stanford.EDU, (650) 723-4138
Kincho H. Law, law@cive.Stanford.EDU, (650) 725-3154
Krishna Sa
raswat, saraswat@ee.Stanford.EDU, (650) 725-3610
Peter Pinsky, pinsky@ce.Stanford.EDU
PROJECT LEADER
Edwin C. Kan,
kan@gloworm.Stanford.EDU, (415) 723-9796
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">
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:
- 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.
- 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:
The 3D geometry/field server based on the minimal SWR
specification
now provides two boundary-movement schemes: the Huygens' principle
which finds the isometric envelope and the mass conservation
principle which satisfies the physical entropy condition.
For implementation using the level-set method, these two principles
only differ in local modification of the level-set function and
can share most of the other functions. User selection of these
principles will be passed through generic attributes of the base
class, similar to the other user-specified parameters and flags.
The level-set boundary movement method has been calibrated in the
etching/deposition simulator SPEEDIE 3.0 for a wide range of processes
including Low Pressure Chemical Vapor Deposition (LPCVD), Plasma
Enhanced Chemical Vapor Deposition (PECVD), Physical Vapor Deposition
(PVD), and Self-Aligned Contacts (SAC) by matching with the 2D
cross section of the SEM pictures, and stringer formation by
qualitatively comparing the 3D SEM pictures. Elimination of non-physical,
rule-base boundary movement has made calibration much easier since numerical
errors can be better controlled. Important physical features, such as
polymer accumulation in high density plasma oxide etching (HDPOE) and
facet formation in the sputtering etchback, are successfully
demonstrated within the new SPEEDIE.
The finite element discretization for the diffusion equation may
encounter numerical undershoot problems, where concentration becomes
negative, if the mass matrix, even though positive-definite, contains
negative elements operating on very large concentration gradients.
Since negative concentration is not only unphysical, but also may
cause numerical problems in reaction and recombination terms.
To guarantee that concentration remains positive for all time
steps, the discretization has to satisfy the maximum principle.
We have analytically derived the criteria for the maximum principle
in the diffusion equation in 1D, which can be extended to 2D and 3D.
Also, to accelerate convergence rate in the initial phase of nonlinear
Newton iterations, damped Newton methods with different line search
schemes are implemented to avoid numerical overshoot. For the new
physical models defined in ALAMODE, the Jacobian matrix may be
very stiff, such as in the case of the five-species model where
dopants and dopant-defect pairs have very different scales of transport
coefficients, and the damped Newton scheme can help the overall
convergence of the numerical system.
The functions to transfer data representations between SUPREM-ALAMODE
and CAMINO are implemented. 3D multi-region test structures have
been validated with the fermi diffusion model. The Oct-tree based
tetrahedral mesh contains anisotropic refinement capabilities and
can take general boundary-representation geometry. A hierarchy of
error estimators, including implicit/explicit and inter-/intra-element
methods, are formulated and partially connected to ALAMODE. Error
fields for explicit inter-/intra-element contributions are verified
for the diffusion equation. Along with the error estimator implementation,
access to strong forms of operators are built, which will later be
extended to Galerkin-Least-Square (GLS) implementation.
In collaboration with Dr. B. Froehlich of the Center for Integrated
Systems of Stanford University, a prototype of interactive visualization
of 3D geometry objects generated by VIP-3D
has been implemented on
the SGI virtual reality systems. Computer-tracked stereo glasses and
gloves can allow users to view objects from different angles and distances,
move objects around and investigate flow lines by injecting particles.
RECENT ACCOMPLISHMENTS:
- The 3D geometry/field server based on the minimal SWR
specification provides both the Huygens' principle and mass
conservation for boundary movement schemes. Parameters and flags
for function calls will be passed through attributes in the base class.
- The level set method has been calibrated in the etching/deposition
simulator SPEEDIE for a wide range of processes including LPCVD,
PECVD, PVD, SAC and stringer formation. The simulation results
successfully show important physical features such as sidewall
polymer accumulation and facet formation.
- Numerical undershoot at sharp gradients which may cause nonphysical
field values such as negative concentration has been identified
with the maximum principles in the finite-element method. Damped Newton
methods have been added in SUPREM-ALAMODE to accelerate convergence.
- SUPREM-ALAMODE is connected with CAMINO for multi-region 3D simulation.
Explicit error estimators with inter- and intra-element contributions
are implemented in ALAMODE.
- Interactive visualization of 3D geometry objects generated by
VIP-3D
has been implemented on the SGI virtual reality systems.
FY-`96 PLANS:
- Test 3D Thermal Module and integration of parallel
solver technology.
- Implementation of 3D Oxidation within Thermal Module
and transition into unified capabilities (both diffusion
and oxidation).
- Implementation of complete server for 3D Deposition/Etching
Module, including 3D parallel solver to support flux calculations.
- 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:
Drs. Paco Leon and Martin Giles from Intel TCAD group were invited
to our weekly SPRINT-CAD meeting on Oct. 18, 1995. The focus of
discussions has been on how to transfer TCAD developmental results
from Stanford to industry and research labs. The traditional method
of transferring software from university to vendors and then to users
may not be appropriate owing to: 1) the new SPRINT-CAD modules are
flexible in physical definition and,therefore, users can switch
models and parameters easily, 2) the new modules are developed within
an open simulation framework, and heterogeneous tools can run under
the same environment and 3) the turn-around time from model and
algorithm development to vendor implementation is too long.
University, research labs, vendors and industry users need
to adapt their roles with the new software development paradigm
and innovative information communication.
Edwin C. Kan
kan@gloworm.stanford.edu
201 AEL, Stanford University, Stanford, CA 94305
Office: (415)723-9796
Fax: (415)725-7731
Date prepared: 1/27/96