Nice reminder of the reasons things go wrong in manufacture, from
shock and vibration to thermal cycling.
But this is now essential tech for "end users" too, to assure supply
chain provenance from tampering and implanting of malicious VLSI.
I first heard of the work of Andrew "bunnie" Huang [0] here on HN, and
the ability to inspect with near-IR radio instead of more difficult
and dangerous X-rays.
I look forward to the day I can borrow or hire an electronics "hand
scanner" microscope and verify that Intel hasn't quietly popped
another backdoor "management engine" into the board I bought.
The equipment is probably the less influential aspect of the trade. You need the skill. You can also rent that by hiring an expert. Good equipment (microscopes, cameras, cars) don’t get you much closer to the goal unless you already know what you’re doing and that can’t be too common, especially in topics like this one.
Good point. Even though I have hardware knowledge and studied VLSI
I'd be completely lost with the raw data.
How I imagine this working - Huang and Applebaum both suggested
similar - is that someone would publish "ground truth" templates of
what hardware should look like. The validation would only flag
suspicious differences (using "AI"/pattern-matching etc) and then
those would definitely be escalated to an expert.
X-Rays are one thing, electron beams are the other fascinating thing. They are interesting, because they are not only used to observe, but also to modify prototypes and fix bugs.
The modified circuits are not very stable, so it can only be used to verify a fix.
(I learned this 20 years ago, not sure if it still a thing.)
You are correct. You can do mask edits or short runs in some cases beam.
Even more interesting is mask level overlay and CD correction using ultrafast absorption based stress fields. There are some heat technologies out there
Still used and very useful for test and debug; I worked with Kleindeik Nanotechnik on the process for the Pi 5's little SoC and learned a lot on the way; a little in this blog post but more in the video I linked in it: https://www.jeffgeerling.com/blog/2024/die-shots-and-transis...
der8auer on YouTube also did an even more in depth video on the nanoprobing, at least, and there are a couple places showing the process of etching with electron beam, but not with a very good explanation.
The benefit of X-ray is they go through everything (so you can see metal). The problem is they go through everything, so you have little signal.
For metal/bump voids, hard to inspect any other way
The harder problem for X-ray inspection is that 1) resolution is difficult to scale economically (possible, but hard). And 2) if any version of wafer or die direct bond dominates (I tend to believe it will), then x-ray won't be good enough anyway, so hard to spend the investment on this.
Experience has shown that a number of people win repeated lottery scratch-off jackpots with a statistically impossible frequency. My hypothesis: After-hours use of semi X-ray inspection equipment.
13 comments
[ 0.23 ms ] story [ 151 ms ] threadBut this is now essential tech for "end users" too, to assure supply chain provenance from tampering and implanting of malicious VLSI.
I first heard of the work of Andrew "bunnie" Huang [0] here on HN, and the ability to inspect with near-IR radio instead of more difficult and dangerous X-rays.
I look forward to the day I can borrow or hire an electronics "hand scanner" microscope and verify that Intel hasn't quietly popped another backdoor "management engine" into the board I bought.
[0] https://en.wikipedia.org/wiki/Andrew_Huang_(hacker)
How I imagine this working - Huang and Applebaum both suggested similar - is that someone would publish "ground truth" templates of what hardware should look like. The validation would only flag suspicious differences (using "AI"/pattern-matching etc) and then those would definitely be escalated to an expert.
The modified circuits are not very stable, so it can only be used to verify a fix.
(I learned this 20 years ago, not sure if it still a thing.)
Even more interesting is mask level overlay and CD correction using ultrafast absorption based stress fields. There are some heat technologies out there
der8auer on YouTube also did an even more in depth video on the nanoprobing, at least, and there are a couple places showing the process of etching with electron beam, but not with a very good explanation.
For metal/bump voids, hard to inspect any other way
The harder problem for X-ray inspection is that 1) resolution is difficult to scale economically (possible, but hard). And 2) if any version of wafer or die direct bond dominates (I tend to believe it will), then x-ray won't be good enough anyway, so hard to spend the investment on this.