Live analysis decaps often mill the cavity first to reduce processing time and better control where the acid flows.

Procedure:

1. Ideally you'd x-ray the package to find out where the die is. Alternatively make a guess or sacrifice an identical chip (ie from completely removing it from the package)
2. Use a micrometer or similar tool to measure from the edges and score out a rectangle where the die should be
3. Use a sharp flat tip bottom endmill to slowly cut down and create a pocket until just before the bond wires (or further if you don't care about taking it live)

Tips:

• Consider creeping in a few mm past where the die actually is. As you etch down it will eat out towards the sides
• If you have to apply any noticeable pressure your endmill is not sharp enough. In this case it tends to creep downwards to give an uneven cut and may be difficult to use
• As a rule if thumb I cut down about a quarter of the package height

Several groups have success with simply sanding the chip down using a Dremel or sandpaper.

Procedure

1. Glue or otherwise secure chip package to something. A ZIF socket also worked with mild success for PDIP. A specially constructed, heavy duty type might work well if this was done often enough.
2. Use sandpaper to get as close to the die as possible. Closeness can be gauged by observing the arc of the bond wires. They may be more easily observed by wetting the chip or using moderate (~10X) magnification.
3. Polishing using a Dremel with polishing paste. It is an open question if any pastes are better or should be avoided.

Advantages

• Readily available materials
• Can be tightly controlled

Disadvantages

• Higher degree of skill required for decent results
• Labor intensive
• Not suitable for live analysis (bond wires are destroyed in the process)