8.3.2 The PIF Geometry Format

The PIF geometry is usually created using the PIF editor PED, although it is possible to write an ASCII PIF file with a text editor and convert it into binary form using PBFM. The following definitions apply to VLSICAP geometries:

Vertex, Point
A pair of cartesian coordinates in two-dimensional space. The origin of the coordinate system and the unit of length are user-definable.

Polygon
A list of vertices connected by straight lines. The train of edges must not be closed. A polygon also has an orientation defined by the order of the vertices. Positive orientation means that the first vertex on the list is the beginning of the polygon, the last vertex the end (vice versa for negative orientation). Polygons are numbered for easy reference. The sign of the number expresses the orientation.

Segment, Domain
A list of at least two polygons. The according polygons are linked to form the closed domain boundary.

Simulation region
The set of all segments in the geometry.

• Different polygons may only share starting/ending vertices but no inner vertices. If a vertex is on more than two polygons they must be split at .

• A polygon must not intersect any other polygon.

• The ending point of a polygon must be identical with the starting point of its successor in the list.

• The ending point of the last polygon must be identical with the starting point of the first polygon.

• The segment interior is located on the left hand side if the boundary is followed in the order specified by the list.

• A polygon on the perimeter of the simulation region (outer polygon) must appear in exactly one segment definition. The rest of the polygons (interfaces, inner polygons) must appear in the definition of exactly two segments.

• Although MESHCP can cope with holes in the simulation region (i.e. an interior conductor consisting of ``Conductor'' material) being specified as a separate face with reverse orientation, the VISTA wafer state representation does not allow this (and therefore PED cannot handle this case correctly). Interior conductors are specified by cutting the face (segment) from the outer boundary to its inner hole, and thus defining it as one face which touches itself on the cut line. This is exactly what MESHCP would do when encountering a multiply connected domain.

• Regarding segments, VLSICAP understands just one face per segment, so each segment with a distinct material has to be exactly one face.

VLSICAP needs boundary definitions where bias voltages are applied to a device. This is a contradiction to the wafer state description employed by the VISTA SFC [Hala94]. Integrating VLSICAP into the SFC controller would need a conversion of all conducting segments to boundaries of the dielectric or semiconductor segments. This is a main obstacle in providing a tighter integration of VLSICAP into the framework.