Chemically removing IC identification??

[gijim]

I was wondering if there was a way to chemically remove the identification off ICs, as a method to prevent (discourage) reverse engineering? I know I could sand it off, but that takes time, and just looks bad. Ideas? I've tried brake fluid and brake cleaner... I'll just keep throwing things at it until it works, unless I get other suggestions. Thanks!

 

[Skogsgurra]

This question pops up now and then. It is often put by someone that has invented a truly amazing circuit that no one has invented before.

Removing IC numbers will not protect a circuit from being reverse engineered. It is relatively easy to find out what class of function it performs and then it is only a matter of probing the pins to see what is going on and in what circuit number. There are even IC testers that automatically identify the IC.

So, if you really think that your circuit is very valuable and unique, you shall apply for a patent. Or, which is a lot cheaper, put it in an FPGA or similar package. If it involves analoge circuitry. an A/D and a DSP may help keep your valuable secret from being pirated.

BTW: Do you down-load music, videos, games?

Gunnar Englund

http://www.gke.org

 

[Comcokid]

Many ICs packages are marked using a laser. This results in a identifier that is equivalent to being molded-in the plastic. The only way to remove the numbers is to - sand it off.

Removing the numbers is a good way to keep the general low-budget copycats from reverse-engineering your circuit. In a sense, you only need to put up enough defense to discourage a copycat from copying. However, for a larger or better equipped reverse-engineering effort, lack of identifiers only slows them down a little.

Depending upon who you think may want to copy your product, patents may not be a defense either. You have to discover you are being infringed upon, and then bring suit. A very co$$tly proposition for which costs are frequently not recovered. It's the small fly-by-night outfits (low assets) that you might be prosecuting. You can't get blood out of a rock. Large organizations will tresspass on a patent in some cases if the market has potential, and they anticipate they can settle by exchanging patent licenses out-of-court.

 

[IRstuff]

The package markings are essentially irrelevant. Pop the lid or dissolve the plastic package and the IC markings on the chip itself are visible for those who really want to know. Exposing the die takes less than half a day to achieve. If anyone is sufficiently serious, they can pay to get any IC reverse-engineered.

The only remotely possible solution is to create a package that performs some essential integral function that gets destroyed upon the processing used for reverse engineering. One such possibility is to embed critical interconnect in the package. A plastic package with embedded, doped polymer interconnects might be challenging to reverse engineer. Even then, it's unlikely that someone who is sufficiently motivated, can't reverse engineer it.

The Soviets famously reverse engineered the 8080 microprocessor in the late 70's. The story was that they were unable to do a particular part of the processor and had to come up with an alternative that showed up as a massive change to the IC layout.

There was a thread about this subject a couple years ago; I think it was in the semiconductor engineering forum, but I'm not positive about that.

Now, one reasonably robust approach is to use embedded flash memory that's locked from external readout. This would require someone to have relatively expensive e-beam probing capability to reverse engineer.

TTFN

 

[itsmoked]

I agree with Comcokid fully, well said. Certainly there is a cost to reverse engineer something. The point is to make the cost of the reverse exceed the cost of asking someone like me to just design the thing from scratch. Most things wouldn't cost that much to design from scratch.

I often remove IC labels, most often when dealing with flaky customers. If I can even vaguely guess they are unscrupulous, say they screw with payment terms, any protos handed to them are erased. I also don't care what it looks like! In some cases I want them to actually realize my intent and "get the picture". I once had a scummy customer who wanted schematics because they "couldn't read the packages". They got nothing as the original agreement gave them no rights to the design, they weren't supposed to even care because we were going supply the product. (They weren't paying correctly.)

In these quick and dirty cases the Dremel tool works fine.
For production there are many places that can do it for about $0.10 per device. They use marking lasers to do it. It looks entirely factory and they can even change the labeling for you. You hand them a tube of parts and they hand them back the next day.

I don't remember the name of the place near me but I'd look it up if you want me too. They gave me a business card that was a laser engraved piece of stainless sheetmetal. I remember because it was kinda scary to handle (sharp).

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[gijim]

Thanks for the input. The circuit isn't all the valuable, I just want to prevent the low budget hobbiests from trying to copy my device and selling it for less. Obviously most companies with dedicated teams can RE it. If they want to spend $10,000 to break it down, that's their choice. I can't stop them.

So I guess the general concensus is that it's a laser etching? Would I need something nasty like nitric acid to remove it? What dissolves the plastic?

 

[IRstuff]

You should consider a simple Dremel tooling. Etching a nonuniform surface poses some difficulties, since the laser damaged portion will etch faster than the surrounding areas.


TTFN

 

[IRstuff]

Alternately, you could consider using a laser marking tool to scrub the part ID

TTFN

 

[itsmoked]

The company I mentioned etched a pocket knife while I was standing there, (in appreciation of my visit), so they could etch an assembled board though it might cost more.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[MikeHalloran]

Inexpensive IC packages are thermosetting plastic.

No solvent will dissolve them.



Mike Halloran
Pembroke Pines, FL, USA

 

[gijim]

I was afraid of that

 

[VEBill]

Don't just remove info. Replace it with false info. About 100x more effective. 'Art of War' and all that.

 

[ScottyUK]

If it is a low cost, high volume product, chip & wire construction with an epoxy 'glob-top' becomes a fair possibility. Sure, you can dissolve the epoxy with hot methylene chloride if you are determined enough and then inspect the bare die, but glob-tops make it difficult to get at the die without wrecking it.

----------------------------------

I don't suffer from insanity. I enjoy it...

 

[itsmoked]

VE1BLL's suggestion if it can be pulled off (not real easy) would really throw a wrench in things.

Mike; I recently visited a company that makes a tool that is an automated IC package remover. (that's all they make)

You drop the IC into a little holder, close the lid, press the button, and it bathes the IC with any number of heated acids.. The plastic all leaves, thermosetting or not. You can dial a depth even. They say it's ostensibly for depackaging for "failure analysis". And I say, "suuuuuure it is"...

http://www.nisene.com/jetetch.shtml

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[ScottyUK]

Interesting machine. I bet they have a really large client base in China! If someone is prepared to go to those lengths glob-tops are out too.


----------------------------------

I don't suffer from insanity. I enjoy it...

 

[IRstuff]

We used to have such a machince specifically for failure analysis, particularly ESD zaps. The first step to a full-blown destructive analysis is the package removal. This is followed by various etchbacks that eventually result in the complete removal of everything but the silicon. From there SEM imaging of the failed input or output will reveal a pit in the silicon where the current went during the zap.

Obviously, the identical processing results in a layer-by-layer dissection of the part for reverse engineering. However, the market for such a machine for this purpose is actually quite limited. It used to cost about $100K to reverse a complex part, so most people who are interested go to house that specialize in this sort of thing. The package removal machine is probably the cheapest part of a more expensive lab that's equipped to do this, e.g., other etchback machines, SEM, CNC-microscopes, mass spectrometers, CAD programs, etc.

TTFN

 

[OperaHouse]

A company I used to work for had this big secret that they had to hide (two silicon diodes back to back and they put it in a DIP header with a cover and some epoxy. Later I upgraded the design to two full wave mini dip bridges. Thede were placed in sockets and the 16 pin dip header was just glued on top of that. The appearance looked quite nice. The point is, glueing a cap on an IC looks quite nice, is cheap, quick, and will fool a few people.

 

[MikeHalloran]

Of course there's a Chinese market for depackaging machines.

They'll import ... one.



Mike Halloran
Pembroke Pines, FL, USA

 

[itsmoked]

Boy....ain't that the truth!

Had a customer who sent a 15.2 million dollar machine to Asia. The techs were told they could not be in the building during 9-12pm one night. They left at 8:30pm. After remembering a forgotten tool they returned at 9:30pm (after dinner) to pick it up. They were shocked to find over 30 people crawling over the partially disassembled machine with cameras, micrometers, huge two handed calipers, and an army of clip boardists..

Really disgusting.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[itsmoked]

Wait Mike.. Wouldn't they need to buy two? lol

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com