SPRINTCAD QUARTERLY SUMMARY

January 1 - March 31, 1996


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 Saraswat, 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:
  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:

For the next decade, it is still expected that many algorithmic improvements on numerical methods and computational geometry for 3D problems will be proposed. The original design of the minimal SWR specification assumes one type of volume gridder in the entire scope of the server. Although attribute mechanisms can support the possible selection among various gridders, dangerous cast of pointers may be necessary and type consistency may also be questionable. The minimal SWR specification has been revised for a unified 3D geometry class with multiple live gridders. The geometry class also serves as the communication media between different gridders. For 3D TCAD support, gridders based on unstructured tetrahedra (EUCLID from Stanford) and oct-tree (CAMINO from Stanford) have their respective advantages in different applications and are included in the specification. Inclusion of the Eulerian-type gridders, such as the one used in the level-set boundary movement, is still under study.

Calibration of etching and deposition simulation has been mostly performed on 2D geometrical profiles based on infinite-width approximation, not only because 3D simulation tools are much more computationally expensive and not widely available, but also because 3D profile characterization methodologies are limited by insufficient accuracy in positioning 2D cross-section measurements. An L-shaped test structure is chosen to overcome this characterization difficulty by 45-degree angle cuts at various positions. 3D simulation is performed using the physical models in SPEEDIE, whose parameters are calibrated using 2D infinite-width trench simulation, and the level-set method, which can accurately and robustly model complex 3D boundary movement. Not only that 3D effects, such as the flattening and thinning of the bottom coverage going into the corners, can be clearly observed in both measurement and simulation, good match between various 2D SEM and simulation cross sections also shows the physical model and the boundary movement method are accurate in 3D. This is an important starting point for genuine 3D analyses on more complex structures.

The use of 1D parameters or 2D profiles in creating 3D structures through geometry modelers such as VIP-3D has potential limitations to accurately include important 3D effects. For example, 3D LOCOS structures generated in this way do not show the effects of enhanced oxidant diffusion at mask corners. A new capability to integrate simulated 3D surfaces has been used in VIP-3D to generate more accurate wafer structures. We have chosen a quasi-3D LOCOS modeling algorithm, which uses a combination of parameterized 2D analytic bird's beak shape equations and a fully 3D oxidant diffusion simulation based on the boundary element method to model corner effects in local oxidation. 2D and 3D simulations are combined in order to reduce computational cost. This heterogeneous approach reduces overall complexity which mirrors the goal of VIP as a means for rapid prototyping.

A generalized oct-tree mesh generation algorithm enables mesh refinement and de-refinement in different directions at various regions. A vector level control function is computed and indicates the directions for which the refinement will be performed. In a contour based refinement scheme, the level control function indicates the directions as the gradient, while in an error estimator based scheme, it indicates the direction where the error will be maximally reduced. Every octant can be refined in either one, two or three dimensions. After the tree is generated, detailed tetrahedralization algorithms are implemented to match the complex geometry and ensure mesh conformity. Then after each simulation step, the vector level control function is re-calculated according to the new gradient or error, and the mesh is adapted to reflect the changing areas of simulation significance. With the tree structure, interpolation error is also greatly reduced since the meshes before and after the adaptation share many common nodes. Grid quality required by the maximum principle is degraded during the adaptation. New algorithmic solution is currently under investigation.

RECENT ACCOMPLISHMENTS:

FY-`96 PLANS:

TECHNOLOGY TRANSITION:

The SRC SUPREM Review was conducted on Feb. 13-14 in Stanford University. The basic modules for physical 3D process simulation developed under the Sprint-CAD project including dial-an-operator physical formulation (ALAMODE), error estimation, quad/oct-tree based gridder (FOREST/CAMINO), level-set boundary movement and geometry modeler (VIP-3D), have been demonstrated in the integrated SUPREM OO7 open environment, with clear definition of procedural interface for geometry/field servers and unified user interface through tcl/tk. Sprint-CAD, as a related ARPA project to the SRC SUPREM efforts, has received very good feedback on the object-oriented design and functional capabilities from the SRC industrial members. There is growing interest from IBM, Lucent and Intel to port the software developed under the Sprint-CAD project. It is also proposed that ALAMODE be the benchmark platform for the bulk diffusion models developed under the SRC/National Labs CRADA projects.

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: 5/9/96