GPU Die 'Exploding'?!?!

[silkstone]

So I removed the Heatsink from one of my GPUs that has been having issues (frequent crashes when loaded) and I noticed a pit in the die.
I wondered if it was a manufacturing defect, but then I noticed a similar pit on the heatsink that lines up pretty much perfectly with the pit on the Die.
This is the first time I've removed the heatsink from the card and so the only thing that I can think could have caused this was part of the GPU die exploding! The pit on the heatsink isn't insubstantial and it can be clearly felt when running a finger across it.
Has anyone else experienced a problem like this with their GPU?

 

[eidairaman1]

Could have been a defect or pitting corrosion

 

[Toothless]

Dude if your die exploded, there would be no die.

 

[DrCR]

The magic smoke may have tried to breach the encapsulation, unsuccessfully, hence the crater.

 

[joemama]

No die exploding found, disappointed.

 

[silkstone]

Could have been a defect or pitting corrosion

On both the gpu and heatsink?
The only thing that was in contact was thermal paste I'm not sure how corrosion would have happened.
Besides the hole in the die is not smooth, it's like part of it popped off.
I've had who cards die on showers of sparks before, but never the gpu itself.

 

[Caring1]

On both the gpu and heatsink?
The only thing that was in contact was thermal paste I'm not sure how corrosion would have happened.
Besides the hole in the die is not smooth, it's like part of it popped off.
I've had who cards die on showers of sparks before, but never the gpu itself.

Looks like a drop of water (sweat) possibly got caught in the thermal paste and became super heated causing a small "explosion"

 

[Bomby569]

it's actually very hard to get that kind of damage with paste and a cooler

 

[Massdeth]

I agree with bombay569. The silicon on processors is very hard stuff, but to get a divot in there like the OP has there would have to have been a foreign object caught in that spot and then the heatsink would be cranked down on it enough to chip it. Being the heatsink has something similar in it that just screams foreign body got in your application when your putting the TIM on the cpu. Did you put it together on the rug? If so, dont ever do that... that is just one of the MANY problems that comes with doing that. Being that the hole looks tapered like it is, it reminds me of what a drill bit does. This is def a first for myself and ive built alot of pc's from the cheapest garbage to really nice customs over 20 years straight.

 

[cst1992]

When I read "exploding" I thought you ran a thousand watts through the die or something.
But then if you'd done that, the VRMs would have popped alongwith the die.

 

[freeagent]

One heck of a hard spot in that TIM

Something must have been present when the two surfaces mated.

 

[Vya Domus]

That's not the "die", it's a top layer over it.

 

[DR4G00N]

it's actually very hard to get that kind of damage with paste and a cooler

I've had it happen with an XFX HD 7950 I had, the vapor chamber on the (defective) cooler expanded from the heat and crushed the die to pieces.

What the OP is seeing is likely a blowout from an internal short, must've happened in a non-critical area though since it still does something but it's toast.

 

[TheoneandonlyMrK]

That's not the "die", it's a top layer over it.

True , but that small portion is not going to be easy to keep cool, that's a hotspot cause, and a half.

Looks like something exited the chip package rather than got pushed into it too, very odd , never seen this like this before.

 

[silkstone]

I agree with bombay569. The silicon on processors is very hard stuff, but to get a divot in there like the OP has there would have to have been a foreign object caught in that spot and then the heatsink would be cranked down on it enough to chip it. Being the heatsink has something similar in it that just screams foreign body got in your application when your putting the TIM on the cpu. Did you put it together on the rug? If so, dont ever do that... that is just one of the MANY problems that comes with doing that. Being that the hole looks tapered like it is, it reminds me of what a drill bit does. This is def a first for myself and ive built alot of pc's from the cheapest garbage to really nice customs over 20 years straight.

These are photos taken after i very first removed the HSF array to re-paste. The paste has never been cleaned before because I was saving the warranty. There was a sticker and I'm non-US.

One heck of a hard spot in that TIM

Something must have been present when the two surfaces mated.

There was nothing that I noticed when wiping it clean. Not that I couldn't have missed it, but the card is about 3 years old and has only been experiencing issues the past few months.

 

[eidairaman1]

On both the gpu and heatsink?
The only thing that was in contact was thermal paste I'm not sure how corrosion would have happened.
Besides the hole in the die is not smooth, it's like part of it popped off.
I've had who cards die on showers of sparks before, but never the gpu itself.

All it takes is a electrolyte to start the process, moisture.

My profession as an Aircraft Mechanic (A&P) sees that kind of corrosion from time to time, it is either rough like that or smooth. Either way the die is damaged and needs replacement.

 

[Nike_486DX]

Looks like corrosion from the liquid that got out of that thermal pipe (manufacturing defect maybe), i had seen this on a couple of graphics cards that feature direct contact heatpipes.

 

[eidairaman1]

Looks like corrosion from the liquid that got out of that thermal pipe (manufacturing defect maybe), i had seen this on a couple of graphics cards that feature direct contact heatpipes.

Interesting you are the other to mention corrosion lol

 

[thesmokingman]

One heck of a hard spot in that TIM

Something must have been present when the two surfaces mated.

There be diamonds!