Seen on a number of high performance, hot running chips to improve conductivity. Nitric acid dissolves both silver and copper and. Mechanical abrasion should also work. It seems it may often be that the die substrate is bonded to the copper/silver heatsink. The actual die is covered by a protective layer of epoxy;. As such, if nitric acid is of limited availability, a cheaper method would be to use hydrochloric acid and hydrogen peroxide as these chemicals are much more readily available. I had mild success with this technique on such a sample, but found that care should be taken if the packaging is multi layered. It should peal apart after some corrosion, so should require more attention (or lots more time), but should not be a major issue. Using over the counter 3% H2O2, expect a moderately sized (couple square inches) sample to take several hours. Flowing/agitated liquid should help considerably as fresh H2O2 needs to be constantly circulated to the copper areas to be effective.

Sometimes the metal can be peeled away when strongly heated. I've also had some luck rapidly cooling it for thermal shock release.

One time for kicks I used the package as an electrode. Seemed to work okay but was slow to eat away.