Our first test case was a PVD deposition of a Ta barrier into a simple dual Damascene interconnect structure. We looked at a 1 micron long section of a 0.4 x 0.4 micron line with a 0.4 micron deep, 0.2 micron diameter via as shown here:

Our code modeled ballistic transport from a source volume above the feature, calculating 3D viewfactors along the way, and integrating the motion of the surface using PLENTE. We virtually deposited 140 nm of Ta on the flats, resulting in the structure below. (The semitransparent yellow indicates the starting structure.)

Due to the shadowing down in the via hole, thicknesses at the bottom of the via are much less (less than 15 nm at depth more than 0.20 microns from the top of the via) than top of the via, causing "bottlenecking" as highlighted in the structure below and the following film-thickness map.



These simulations took about 45 minutes to run on a 2.2 GHz Pentium 4 linux box.