Charge to the NSF Workshop on Structured Design Methods for MEMS
Dr. Bernard Chern
Microelectronic Information Processing Systems (MIPS)
National Science Foundation
4201 Wilson Blvd.
Arlington, VA 22230
tel: 703/306-1940, Fax: 703/306-0610
The goal of research in rapid prototyping is to develop and integrate
methodologies, tools, environments and technologies needed to be able to
automate the rapid, efficient and accurate design and construction of
processes, artifacts and systems of artifacts. A key long term objective
is to develop a design methodology that can be applied generally to
mechanical and electromechanical systems.
This workshop is intended to identify and encourage research efforts on
implementing MEMS design methodologies. The scope of these efforts,
however, includes not just the design methodology itself, but also the
design tools, design environments and design technologies that will be
available for rapid prototyping using MEMS.
At this workshop we will grapple with research needed to create a VLSI-like
design methodology (including tools, environments, levels of abstraction,
etc.) for the MEMS technologies in which there is a clean separation between
design and fabrication; that is, between the design community with its
concerns about CAD tools, design environments, etc., and the fabrication
community with its concerns about equipment and processing capabilities as
well as such customer servicing criteria as cost and responsiveness.
To achieve such an outcome we will need to answer some key questions:
- Can we use the same generic layering model for all MEMS fabrication
processes? Should it differ from "the most generic model" described in
C.A. Mead's Preface to the Workshop on
New Paradigms for Manufacturing?
- Can we identify a common digital specification language that can be
used generally to describe the desired prototype in terms of the resulting
geometry on each layer? If the answer is CIF (Caltech Intermediate
Format), is CIF adequate for present and future systems?
- Can we use this digital interface (consisting of CIF, the generic
model of layered fabrication and the common understanding of the standard
process) to achieve a clean separation between the design and processing
activities? (See R. Sproull's paper: "Digital Interfaces to Fabrication"
from the Workshop on New Paradigms for Manufacturing.)
- What benefits could be derived from the use of 3D modeling in the
MEMS design methodology? In SFF, design is done using 3D modeling (SIF -
Solid Interchange Format). The 3D model is then sliced yielding data in
L-SIF form (L-SIF - Layered SIF). Is the final physical shape part of the
MEMS design process or is the design carried out in terms of function and
2D descriptions in CIF? How does the MEMS design methodology relate to
that for SFF? (See paper by C.H. Sequin and S. McMains titled "What can we
Learn from the VLSI Revolution?" from the Workshop on
SFF Design Methodologies.)
- Can we make the design tool hierarchy independent of process
evolution (smaller feature sizes, thinner polysilicon layers, etc.)? In
the VLSI domain the fabricators are driven by the design community through
interaction across the "clean separation" interface; can this be done in
- What levels of abstraction are appropriate to MEMS mechanical
design and how can we improve the ease of moving among these levels?
- Does the current design methodology used in MEMS enable a
compatible treatment of the electronic and mechanical structures in the
MEMS system? Would such a treatment be advantageous?
- Does today's MEMS design methodology start with function or with
shape? SFF design starts with shape (3D model) but VLSI starts with
function (HDL description). Which approach is best, in MEMS applications,
for the designer, the fabricator, the user?
- How much commonality is possible between MEMS design and SFF
design: language? interface? levels of abstraction? design tools? etc.
- The MCNC/MUMPS infrastructure has been successful in establishing a
standard process and making it widely available. What can be done through
improvement of the MEMS design methodology to enhance the potential for
rapid, error-free product generation within the present process capability?
A successful workshop for developing a common design methodology for the
MEMS technologies will result in:
- A strong argument for an NSF/MIPS research program focused on
systems design to make for the rapid prototyping of MEMS structures.
- Definition of a common low-level layer-based digital interface
descriptive language for present and future MEMS technology
implementations. This can be a great advantage, making a variety of MEMS
technologies available to the designer (without learning a new interface
language) and making many customers available to the MEMS fabricator
without the need to invest time and energy bringing the designer up to
speed to address his technology. Further, modifications and advances in
the MEMS process area could be accommodated within the same design
framework, merely involving changes in the design rules. Also the digital
interface language would permit design submission over a network so that
brokerage services, such as MOSIS and MCNC, become practical.
- Steady accumulation in the industry of feature and object
description libraries, in a common language, that can be incorporated in
the design heritage of the field and that will encourage the refinement of
a hierarchical design methodology that will make them useful to the entire
- Momentum to create software paths from the higher level
descriptions (with which the designer starts) to this digital interface
- Momentum to create paths (algorithms and translation code) to go
from this common digital interface language to the languages that drive
the MEMS fabrication systems.
- Refocusing by those active in MEMS-based prototyping from the nuts
and bolts of the design process to the optimization of system
implementation along the dimensions of time to market, multiple sourcing,
reuse of design building blocks, de-skilling of low level design activities,
- Lower cost, reduced delay and fewer errors for system
implementation using MEMS-based rapid prototyping technology. (Lower cost
because the design is cheaper and more designs go down this design path to
many competitive vendors, reduced delay and errors because the design path
is significantly automated and takes advantage of design heritage)
- Increased exploration of sophisticated MEMS product design
alternatives because the time and cost for experimental implementation are
brought within reach so that there will be more cycles of learning about
market requirements incorporated in each product generation
- These techniques for using a low level digital descriptive
interface in MEMS can (potentially) and will (very likely) be translated
into enhancements for more general mechanical system design
where they will provide an impetus for extending design automation to
ever-widening regions of mechanical design space
Previous: Structured Design Methods for MEMS
Next: Workshop Discussion Group Members
Return to NSF MEMS Design Workshop Index