EPROM/EEPROM/flash can be erased by shining UV light or x-rays into the surrounding oxide. This slightly oxidizes it, letting the floating gate to drain out.

Usually a masking compound is applied to selectively block UV to the firmware, with red nail polish being the most popular or at least traditional.

McMaster

1. https://www.amazon.com/gp/product/B00LT8S30G/
2. Bought at local store a while back and it seems to work well enough
3. Bulk removal with acetone, but leaves some residue around edges
4. Remaining residue softened with aceotne
5. Remove using single hair brush if possible
6. Can be more aggressive with PP monofilament wire
7. Cold H2SO4 also works well. Wash with IPA then acetone
   1. Washing with water may really heat up the mixture

Some chips with shielding against this type of attack can be bypassed by shining light at a sharp angle

Use nail polish or similar compound. Most materials will block UV, so just about anything will work. Nail polish tends to be easy to work with including application and (acetone) removal.

Use a fine point pipette for simple chips.

For finer masks, place the chip in a large bowl with a shallow layer of acetone at the bottom. This creates an acetone “shield gas” to prevent tools from drying out. Then put some nail polish on a single bristle (maybe held by tweezers) and it should not dry out, allowing you to paint it onto the die.

TODO: add links to misc papers

Many systems have expectations that a piece of hardware will complete an operation in so long. For example, hardware may assume that an ADC will always complete within 100 ms. If evens are generated at 200 ms this normally might not be a problem. However, if one can find a way to stall the external hardware internal queues may overflow and result in unexpected states that can leak data or other vulnerabilities.