Laser decapsulation of a semiconductor chip - GIF

[Jon]

Laser decapsulation of a semiconductor chip to determine the genuineness of the die.




Previously:

Laser graffiti remover - GIF
Laser removing oil from engine block - GIF
Laser cutter cuts intricate pattern on cylindrical workpiece - GIF
Micro laser soldering - video

[jdurand]

Reminds me of the long hours dripping hot sulfuric acid on packages so we could see what the competition was making. The most interesting was encapsulated in plastic for a paperweight, someone pocketed at the other company. Got good early info off that once we got into it.

[metric_taper]

The military side of the company I worked for used special conformal coatings to prevent just this sort of espionage of their custom chips. But seems with a laser and good operator feed back, this stripping would thwart that.

[jdurand]

We had a defense against peeping engineers, throughout the chip were fake vias. Instead of a full hole in the vapox we put gate ox. When you plate the aluminum over it you get a dimple in the metal just like it was a real connection, but of course isn't connected to anything. Of course if connected the chip stopped working. You'd have to figure out how the chip worked to deduce where the errors were and at if you could do that you didn't have to peep at our design in the first place.

[metric_taper]

We had a defense against peeping engineers, throughout the chip were fake vias. Instead of a full hole in the vapox we put gate ox. When you plate the aluminum over it you get a dimple in the metal just like it was a real connection, but of course isn't connected to anything. Of course if connected the chip stopped working. You'd have to figure out how the chip worked to deduce where the errors were and at if you could do that you didn't have to peep at our design in the first place.

I bet that was fun to reverse engr. Sneaky bastards!
I only ever worked with programmable logic devices, but career wise didn't do that path, as it seemed shake-bake qual became my calling. I got pretty good at EMC hardening. I figured out methods of designing filters in the boards just using traces and ground planes. The real trick was figuring out how to get a low inductance ground connection to the board planes. As well getting the kids that were designing the circuit cards, to put in proper decoupling to the microprocessor parts to minimize RF radiation.

[jdurand]

I started in shake and bake, was certified for manned space flight, safe & arm, and airplane part certification, then went on a trip to California to learn a new test system they bought. And...the California guys hired me.

After my stay in static memories, CCD arrays, and microprocessors (I used to have the original F8 processor in wire wrap), I did some work at the Palo Alto R&D facility, then moved to hybrid parts for a break. There I just had to maintain the equimpent including the big cutting lasers.

From there I went off to help start the computer sound business where I became really friendly with the MMI programmable logic...no software, just draw a bunch of x's on a drawing of the chip where you wanted fuses blown. I would program spare pins to make somewhat random signals and have those go off to other chips.

At a later company we were showing our latest voice boards at a trade show. We later received a law suit claiming we'd stolen the design from another company. Their engineers had closely examined the board we had on dispaly and verified it was their design. Our lawyer couldn't wait to get to court with the board, I'd had my technician make it up special for the show...not a single IC on the board was the right one. I had him match pin count but that's it, he had some Z80s on there, and assorted TTL, CMOS, and analog chips. If it fit he soldered it on.

We won the case.

I still chuckle about that at times.

[metric_taper]

I used to have the original F8 processor in wire wrap

I take it you worked for Fairchild. I only ever used some of their analog ICs, and 422 transceivers in my designs. We really didn't have a choice in components, as there was a whole group of component engineering that specified parts mostly from fixed sources, Intel, Motorola, National Semi, so I was pretty much stuck with what had been given a part number in any new designs. And they had buying agreements with those sources. Our volume was real low with avionics. You know if you built 200 of some part number that was a big volume. And that was over a 15 year or so period of deliveries. But I worked for the bizjet side, not the commercial jet (air transport division). And it was a major difference in safety, 12 dying vs 100s. But it still was 10e-9 failure per hour fault requirements. But they were designing fail op, for auto land, and we just had the fail safe requirement. That was where I spent more engineering time in the safety side. I started right when Beechcraft was finishing the Starship (a failed design, pusher prop, canard) and it being all carbon fiber, the FAA was engaged in new regulations for Lightning and RF susceptibility worries. So you end up blooming where your planted.
I only heard stories of industrial espionage on our products, from the 1960/70s. It was all analog avionics, old timers told me a story of King radio engineers taking schematics of our autopilot (King started with engineers that left Collins Radio), but we had a servo amplifier design that was extremely safe driving a PM servo motor. The game is torque control, and limiting it precisely. It's also the fault analysis. But everything was designed to prevent a hardover fault event, one that could break the airframe.
I keep a watch on Boeing and their 737max, they really screwed up. They were always known for overkill safety in their designs, vs Scarebus, that had many landings short of the runway. Boeing had a mentality to never take control away from the pilots with automation, (Airbus did just the opposite) clearly MCAS was marketing wanting a plane that they could call the same so no retraining was needed. This can kill their company, as now the FAA is in anal mode. And they don't have enough smart engineers, so they will have to be trained on every aspect of how the plane works, and with under staffing, it will take a very long time for approval. Worse is now the rest of the industry will be under scrutiny. I was a DER just before retiring. I'm glad I'm not there no more, as this will impact every certification for many years until the FAA goes back to sleep. It has invited the certification authorities from around the world as well, so every one of them has to be made satisfied to its safety. Even they don't trust the FAA.

[jdurand]

When I worked in the Shake & Bake industry we were also called as expert witnesses at times. Things like x-ray a TO-3 power transistor and find a TO-92 attached to the pins inside. It would work long enough to pass normal inspection testing (1-2 seconds of on time), only to fail much later when the seller is long gone.

We got pretty good at spotting things like the printing on the parts being too crisp (low use printer), normal parts showed more rush and wear in the printing.

We sometimes stamped the bottom of those parts with UV ink, then returned them as rejects. We'd get a fresh batch shortly with new date codes and sure enough, there was our stamps on the bottom.

I also got to inform a large computer manufacturer that their previous test company faked their test results since the laws of physics forbid any parts from passing one particular test. That caused a "bit" of a storm.

Then I got in trouble for shipping missile parts with all the original markings plus our test stamp. Long lecture how I'm to remove the word MEXICO from the parts as it makes the buyer think these are foreign parts. I tried to point out that Mexico IS a foreign country, but was strongly informed that it is part of the USA.

My wife says she also had to be careful to remove all the INDIA labels from steel they used on government buildings, gave the false impression it was foreign steel when it had only vacationed there.