EVOLVE AMR

EVOLVE AMR
Adaptive Mesh Refinement

Provides increased spatial resolution at locations with high curvature.
Grid density adapts to local mean curvature.
Requires fewer total mesh elements that using uniform mesh.

AMR is demonstrated using deposition on the pyramid shown.

For physics based transport and reaction simulations, surfaces (the "zero level" on a grid) are extracted many times.

Grid density and extraction need to provide reasonable results.

Example: A few nm "deposited" on a ~150 nm pyramid, with 2 levels of refinement ("0 and 1")

The left-most image shows an extraction of the input feature at fairly high grid density, using 2 levels of refinement (2**0 and 2**1). There are a lot of surface elements. The image on the right is a similar extraction after evolving the AMR mesh after performing a deposition simulation.

Example: Same process and grid density as above, NOT using AMR

The 2 figures to the left show extracted nanopyramid surfaces before and after simulated thin film depostion, as in the example above, but NOT using AMR. Note the rough edges that can cause modeling trouble.

Example: Extractions using 4 levels of AMR, at a lower grid density than above.

The images to the left both show extractions of the input feature using 4 levels of refinement (2**0 to 2**3), but with different parameters. The left-most extraction clearly shows all 4 levels, and is very rough. It would not be a good starting point for simulations. The image on the right would be good for process simulations. In addition to different parameters, there was some post extraction smoothing. There are fewer surface triangles than the 2 level AMR results above.

We have successfully used our AMR in simulation studies, but the multiple level AMR is "alpha". We are looking for joint applications projects, or joint (software) tool development projects, to refine our approach.

PE Home About PE Contact PE
EVOLVE AMR EVOLVE MM EVOLVE 3D EVOLVE 3D-ICs PLENTE