The info that Evan posted is a reasonable explanation of one thing that is probably happening. Plastics in general have a much less-controllable coefficient of expansion under changing thermal conditions, and teflon is worse than most (although the information that is posted shows change in microns, and doesn't indicate whether this is microns per mm, or an absolute change on a specific part dimension, or what exactly).
In addition, you are working with a material that is more difficult than most to keep in one place for machining processes. It wouldn't surprise me if the dimensional expansion you are seeing is a combination of the edge of the material backing away from the cutter, and the part slipping slightly on the tape. The tolerances you are attempting to hold are the kind of thing that always put me on notice (in a couple of previous plastic fab lives) that the designer didn't know what he was doing, especially if I saw a +/-.001 tolerance.
As AppliedProto mentioned, is it possible to do any sort of top clamping plate arrangement? My preference would be to do that with an acrylic (Plexiglas)top plate, rather than aluminum, and make it of such a size that you machine both materials at the same time. If you do that with aluminum, you may embed some aluminum in the teflon. It may be obvious, but dead-sharp tools are a must for this material.
How do you know you aren't stretching the part a little when you remove it from the setup? Are you measuring in place?