Voxelbased Heterogenous Modeling for SurfaceMicromachined MEMS - PowerPoint PPT Presentation

1 / 18
About This Presentation
Title:

Voxelbased Heterogenous Modeling for SurfaceMicromachined MEMS

Description:

(Inverse Problem) Given a process and a desired part geometry, what do the masks look like? ... Ananthakrishnan, Radha Sarma for their work, help, and support. ... – PowerPoint PPT presentation

Number of Views:45
Avg rating:3.0/5.0
Slides: 19
Provided by: zeartist
Category:

less

Transcript and Presenter's Notes

Title: Voxelbased Heterogenous Modeling for SurfaceMicromachined MEMS


1
Voxel-based Heterogenous Modeling for
Surface-Micromachined MEMS
  • Andy Perrin
  • Venkat Ananthakrishnan
  • Radha Sarma
  • G. K. Ananthasuresh
  • (NSF Grant DMI 99-70059)

2
Motivation
  • Surface-micromachined MEMS devices can have very
    complicated geometries.

MUMPS micromotor
3
Motivation
  • Two problems
  • (Forward Problem) Given a process and the masks,
    what does the part look like?
  • (Inverse Problem) Given a process and a desired
    part geometry, what do the masks look like?
  • We care about the part geometry, so we want to
    solve the inverse problem, but to solve it, we
    need the forward problem.

4
Forward Problem Strategy
  • Define some mathematical operators that can be
    composed with each other to yield process steps.
  • If each process step can be reduced to some
    composition of operators, it does not matter how
    that step is actually implemented.
  • Develop a practical implementation.

5
State-change operators
Query operators
6
Forward Problem Voxels
  • Voxels are well adapted to
  • Heterogenous modeling (multiple materials)
  • Layered manufacturing processes
  • Creating meshes for later analysis (FEA, etc.)
  • Topology optimization
  • Etching multiple layers
  • Problems with voxels
  • Rendering speed
  • Memory requirements

7
Forward Problem Heterogenous Modeling
  • Voxels can store local material information.
    (Currently, we store a string indicating the
    material in each voxel.)
  • Our mathematical model can be combined with Kumar
    and Duttas (1998) representations of
    heterogenous solid models.

8
Forward Prob Implementation
  • Four layer types
  • Conformal
  • Stack
  • Via
  • Planar
  • Two etch types
  • anisotropic
  • isotropic

9
Kinds of Layers
Stack
Conformal
Via
Planar Via Stack
10
Kinds of Etches Anisotropic
Before
After
11
Kinds of Etches Isotropic
Before
After
12
User Interface
13
Example
'stack' 'Substrate' 0.06666 1 'stack'
'Oxide' 0.13332 2 'etch' 'etch1' 0.13333
'Oxide' 'anisotropic' 1 'conformal'
'Polysilicon' 0.06666 3 'etch' 'etch2'
0.066663 'Polysilicon' 'anisotropic' 2
'etch' 'etch3' 0.19998 'Oxide'
'isotropic' 2
14
Inverse Problem Strategy
  • For each step of the given process do the
    following
  • If the process step is a deposit command, deposit
    that layer in a new intermediate model (IM).
  • If it is an etch command, compare the current IM
    to the given model and extract a mask etch the
    IM with it.
  • Check/correct masks if they need modification
    update IM.

15
Inverse Problem Strategy
  • Mask correction routine yields three
    possibilities
  • No solutions (impossible to build with given
    process)
  • One solution (convenient!)
  • Many solutions (Develop a figure of merit and
    choose the best.)

16
Inverse Problem Limitations
  • Restricted to conformal layers
  • Restricted to anisotropic etches

17
Future Work
  • Incorporate via, planar layers and isotropic etch
    in inverse problem
  • Apply topology optimization and combine with
    inverse problem to automatically design
    structures.
  • Feature based modeling

18
Thanks to...
  • Dr. G.K. Ananthasuresh, Venkat Ananthakrishnan,
    Radha Sarma for their work, help, and support.
  • NSF Grant DMI 99-70059
Write a Comment
User Comments (0)
About PowerShow.com