I was replacing the USB isolator IC on one of my temperature controller boards, and nearly chucked it out when I realised it might be neat to try decapping it. The IC (an Analog Devices ADuM4160) is rather fancy (it contains two USB transceivers plus several isolation transformers) and turned out to be very interesting.
I had previously seen Ben Krasnow's video on decapping, where he used fuming nitric acid. Lacking fuming nitric acid, I just used "normal" conc. acid with the chip placed on a hotplate at 120°C. After a while, this ate into the surface of the package and revealed the three dies. I then got a bit impatient and chucked the whole chip into acid to dissolve everything and release the dies.
The USB transceivers are on separate dies from the transformers - the central die contains five tiny transformers. The coils are 500μm OD, 240μm ID and have 15 turns each, with a "wire" width of only 4μm. The actual dies measure 1.15x1.6mm (small) and 1.3x3.3mm (large).
Here's a video of the process:
I tried getting some photos down the microscope - they didn't turn out too badly, but this was just holding the camera near the eyepiece. I really need to get a decent microscope/camera setup...
The central transformer die is marked with some text, and I can just make out the Analog Devices logo with the letters ADI next to it.
The transformers are really neat - here's a tracing I did of one of the coils, just to make the arrangement a bit clearer (click for larger image):
And here's a diagram from Analog Devices' own paper on their "iCoupler" technology: