News Intel reportedly denies RMA for crashing Core i9-14900K CPU due to liquid metal thermal paste usage — liquid metal erased the markings and serial n...

"Apparently, the liquid metal had corroded the processor's integrated heat spreader (IHS), removing the chip's markings, including the processor model, batch number (FPO), and 2D Matrix (ATPO)."

Not fully true. The directions to use liquid metal means you have to 'scratch' the surface of the IHS. They usually come with a little scrubbing pad to use. That in itself is considered "physical damage". And scratching off the Intel markings that help them identify what the CPU is, batch #, etc, yea, I can easily see Intel saying no for an RMA.
Now, I'm not a scientist, so I may be corrected on this, but liquid metal itself does not cause corrosion. However, it can definitely cause corrosion if you use it in a custom water cooling setup with a copper heatsink and aluminum somewhere in the loop. (I'm not sure if that's just with liquid metal, or for everything. I think it's just for liquid metal though).
 
  • Like
Reactions: TJ Hooker
I'm not really sure why anyone would use liquid metal between the IHS and cooler as the benefit would be very low compared to the risk/disadvantage.
Now, I'm not a scientist, so I may be corrected on this, but liquid metal itself does not cause corrosion. However, it can definitely cause corrosion if you use it in a custom water cooling setup with a copper heatsink and aluminum somewhere in the loop. (I'm not sure if that's just with liquid metal, or for everything. I think it's just for liquid metal though).
It depends on the compound, but liquid metal definitely can cause corrosion over time. Roman (Der8auer) has mentioned this and it seems largely due to chemical reaction with gallium. He talked about it on a video about his retired Threadripper video editing system.
 
It is probably related to galvanic corrosion where the electrical potential of different metals causes corrosion.

Google AI says:
Explanation
When two metals with different corrosion potentials are in electrical contact with a conductive liquid, one metal becomes the anode and corrodes faster, while the other becomes the cathode and corrodes slower.

You can take advantage of this process by creating an ablative faster corroding add on to things such as ship hulls.
 
Show the picture!
The IHS was bent from the ILM and the chip wasn't even getting directly cooled.
It is a funny fail but pretty particular to LGA1700 and should serve as a warning: don't use LM on the outside of your IHS and the factory ILM with LGA 1700. Liquid metal is too thin to fill in the bending gap. A contact frame should be used at a minimum. Although LM on the outside of the heat spreader is losing a lot of it's benefit and adding a lot of risk so I wouldn't personally use it in that situation. Too much risk per reward.

Guy probably thought there was something wrong with the silicon he was barely cooling.
 
That headline was a pretty cheap shot - not used to seeing that sort of thing from you guys. File off/sand off the model number and serial number from all sorts of products from all sorts of companies and they'll deny a warranty.

That said, and given the optics of the situation, I'm surprised Intel didn't cut him some slack, assuming he had some way of showing that what he wanted replaced was the same model and series chip he'd bought - which may not even have been possible.
 
I'm not really sure why anyone would use liquid metal between the IHS and cooler as the benefit would be very low compared to the risk/disadvantage.

It depends on the compound, but liquid metal definitely can cause corrosion over time. Roman (Der8auer) has mentioned this and it seems largely due to chemical reaction with gallium. He talked about it on a video about his retired Threadripper video editing system.
I've been doing it since forever. Depending on the wattage being pulled, the difference can get to ~5C - maybe more on something like a 13900k.

AFAIK you can clean the LM and restore the markings with some chemicals but you need to be very gentle cause those same chemicals can remove the marking by themselves. But yeah, if you care about warranty, stay away from LM, that would be kinda self evident.
 
Show the picture!
The IHS was bent from the ILM and the chip wasn't even getting directly cooled.
It is a funny fail but pretty particular to LGA1700 and should serve as a warning: don't use LM on the outside of your IHS and the factory ILM with LGA 1700. Liquid metal is too thin to fill in the bending gap. A contact frame should be used at a minimum. Although LM on the outside of the heat spreader is losing a lot of it's benefit and adding a lot of risk so I wouldn't personally use it in that situation. Too much risk per reward.

Guy probably thought there was something wrong with the silicon he was barely cooling.
I've been using the stock ILM with liquid metal on 3 different CPUs and 2 different mobos. Haven't had an issue yet. The trick is to spread the liquid metal as thinly as humanly possible.

On the other hand the problem is that the first application will get absorbed from the cooler within a month, so you have to do it again a month later and then you are good to go for a couple of years.
 
That's precisely why even the LM can lower the impact of the IHS bending due to ILM most ppl would go for the contact frames instead, liquid metal corroding the IHS is bad for RMA with understandable reason, cost more and didn't stop the IHS deform over time... a good paste/PTM plus those CF is a much better option.

LM just make sense to me if one use those direct die cooling solutions where you will void the warranty anayway.

For the affected customer, it's a sad RIP for his investment in the TOTL expensive intel which smokes itself, just yet another victim
 
I've been doing it since forever. Depending on the wattage being pulled, the difference can get to ~5C - maybe more on something like a 13900k
Maybe if you're using bad paste, but it's usually going to be around 1-1.5C better than PCM and 2-2.5C better than a good paste. To me that simply isn't worth the disadvantages, but I can certainly see it if delidding to replace TIM or direct die as those always net bigger deltas. Of course when doing direct die there's the big advantage just from removing the IHS so Kryosheet becomes a really compelling option for long term usage.
 
  • Like
Reactions: rluker5
Maybe if you're using bad paste, but it's usually going to be around 1-1.5C better than PCM and 2-2.5C better than a good paste. To me that simply isn't worth the disadvantages, but I can certainly see it if delidding to replace TIM or direct die as those always net bigger deltas. Of course when doing direct die there's the big advantage just from removing the IHS so Kryosheet becomes a really compelling option for long term usage.
By PCM you mean something like the PTM7950? Well sure but that's not traditional paste. Still LM is better when you have a high heat CPU, since conductivity of LM increases with temperature. There is no way i'd be pulling 85c on full CBR23 blast on a 14900k without liquid metal (single tower air cooler).
 
Maybe if you're using bad paste, but it's usually going to be around 1-1.5C better than PCM and 2-2.5C better than a good paste. To me that simply isn't worth the disadvantages, but I can certainly see it if delidding to replace TIM or direct die as those always net bigger deltas. Of course when doing direct die there's the big advantage just from removing the IHS so Kryosheet becomes a really compelling option for long term usage.
If one is using LGA1700 the difference in thermal conductivity of LM vs a good paste will become more as the bending overtime do significantly/noticeably worsen the traditional paste effectiveness, back in the 12700KF era the average of my CPU brand new on a U12A vs 1 year later worsened about 2C even after repasting and the CPU is noticeably bent, when it get an extra 0.5mm of gap the LM could get further 2-3C improvement in delta, but a day 1 CF could remove that altogether
 
I've been using the stock ILM with liquid metal on 3 different CPUs and 2 different mobos. Haven't had an issue yet. The trick is to spread the liquid metal as thinly as humanly possible.

On the other hand the problem is that the first application will get absorbed from the cooler within a month, so you have to do it again a month later and then you are good to go for a couple of years.
You sound like you monitor your PC when doing something with potential routes of failure. Maybe you use the washer mod to decrease ILM pressure on your CPU. I monitor my CPU temp more than I should as I still have an old habit of AB up in the corner in games so I would notice very quickly if I had a cooling issue and that seems common among those going to the lengths of using LM.
I also used to spread LM as thin as possible to keep it from dripping as the surface tension doesn't support much thickness (still do with my old 1150 CPUs, but just under the socket). And it will very slowly absorb into nickel plate and quickly into copper. Even when I painted the copper with it and baked it in for a while. Which could make it thinner and make the occurrence of this (images from Igor's Lab articles):
CPU-03.jpg

lga1700_cpu_convex-2048x1152.jpg

look like this CPU that is the source for this article:

07162014-cc4e73d2c4dfd99e3945eaa-1.jpg

It looks like the copper was soaking up the gallium part of the mixture and leaving the rest indium and tin rich where the copper was contacting the IHS. Raising the melting point and leaving that residue. The middle was probably fine at first, but thin gets thinner and unless you are keeping an eye on it you have a chance of running the CPU under conditions known to destroy them.

Air gaps are bad.

And I haven't done a lot of testing on the outside of the IHS with liquid metal, but even under ideal conditions I expect results similar to what you would see here: https://www.tomshardware.com/best-picks/best-thermal-paste which isn't a lot, but how much those couple of degrees are worth and how much is the cost of playing around with our pc varies per person.

Under the die, if you have traditional TIM and not soldered, then your decreased surface area of contact is more of a constriction on heat removal and combined with the decrease in thickness of the TIM, LM can give you a much larger difference.
 
  • Like
Reactions: TheHerald
Sounds like an excuse to deny warranty, the 2D matrix printed on the outside edge of the processor contains the full serial number (full ATPO), so even if the top is unreadable it should not matter at all. It is printed on two different places, the IHS itself and the substrate, and looks like a rectangular QR code.
 
Last edited:
Sounds like an excuse to deny warranty, the 2D matrix printed on the outside edge of the processor contains the full serial number (full ATPO), so even if the top is unreadable it should not matter at all. It is printed on two different places, the IHS itself and the substrate, and looks like a rectangular QR code.
TBF, for normal warranty situations this sounds like a fair decline as using LM would potentially be conductive, and also for those who dare to use such TIM are likely hardcore overclockers, so considering the chances of pumping excessive voltage manually killing the CPU is a "at your own risk" move in most of the times.

The issue this time around is that the RPL SKUs are known to be self frying, so in normal circumstances, for brand image protection one would expect intel to honor the RMA.
 
Intel wasn't wrong in general practice terms, but this debacle has already been bad PR for Intel, and denying one like this still has bad optics as said previously. Indeed some make the calculated risk and take it, but that calculation couldn't factor in that these CPUs would be faulty. It erodes confidence and trust in both Intel's products and their customer relations.

What do you folks think this person is going to do? Drop in a new 13900K or 14900K in the existing mobo and call it a day or have a chip on his/her shoulder and jump ship to Intel's competition?

That's the difference between "the customer is always right" and blaming the customer (remember Newegg having this problem?) Being transparent though, I usually agree with Louis Rossman, so that tells you something...