Bare die mount susceptible to TIM pump-out?(Now: crazy plot for bare die IX)

[Ferzerp]

I'm tired of replacing the TIM under the IHS on my 3770k because of TIM failure under it. Does anyone know if a bare die mount can suffer the same effects?

 

[taltamir]

What is TIM pumpout?
What TIM are you using?

 

[Ferzerp]

TIM pumpout is when repeated heat cycling eventually causes the TIM to migrate and degrade in performance over time.

In my case, it is a difference of around 15C once it gets bad. It happens under my IHS with NT-H1 pretty badly.


TIM choices I have:
AS5
Antec Formula 5 (no clue where this came from)
NT-H1
1 more application of IX.

I am tempted to somehow figure out how to make the IX work on the bare die (maybe with a shim to insulate the non-die portion of the cpu).

I don't know if I'm that brave. I do have somewhat of a lemon of a 3770k, so it wouldn't be a huge loss if I destroyed it in the attempt. I would basically have to mask off everything but the die estimate the amount needed, plop it on to the center of the die, and go from there.

That seems risky to me, but potentially interesting.

edit: maybe high-temp caulk?

 

[Ferzerp]

I'm thinking maybe something like this... Maybe caulk or "instant gasket" which would be in a smaller tube and be easier to apply. Also, the rubbery-ness of it would allow it to compress a bit if it ended up higher than the die.

issues:

What will instant gasket solvent do to the cpu while curing?
What will excess melted IX do to the non-die portion of the CPU while it's hot?
Spring mount for H100, or leave my slightly modified screw down mount (will have to check that it will screw down far enough)?
What will IX do to the die?

 

[ShintaiDK]

Could just be IHS to DIE height related instead of TIM. Unfortunately you will first know what of the 2 reasons it is when you delid.

 

[Ferzerp]

Could just be IHS to DIE height related instead of TIM. Unfortunately you will first know what of the 2 reasons it is when you delid.

My IHS sits directly on my die. One of the first things I did was get rid of the excess rubber adhesive, and sand down the height of the IHS "legs" (ok, not legs since it's a solid square, but that explains it) so that it is really just a plate that sits on the die.

 

[Idontcare]

I'm tired of replacing the TIM under the IHS on my 3770k because of TIM failure under it. Does anyone know if a bare die mount can suffer the same effects?

Before attempting something irreversible like IX on your bare-die (which will most assuredly rip-out the die if you attempt to unmount), I recommend you first try liquid ultra or liquid pro and see if they give you similar pump-out issues.

 

[Ferzerp]

(which will most assuredly rip-out the die if you attempt to unmount)

Even if I do it above the melting point of IX? 😉

liquid ultra or liquid pro under the IHS, or bare die? Also, there is far less mad scientist factor to that, so it's not quite as interesting!

 

[ShintaiDK]

My IHS sits directly on my die. One of the first things I did was get rid of the excess rubber adhesive, and sand down the height of the IHS "legs" (ok, not legs since it's a solid square, but that explains it) so that it is really just a plate that sits on the die.

Arh ok, so you delidded, worked the IHS and replaced TIM and now having a problem? Sorry I had the idea that it was "default" 🙂

 

[nikosa43]

Hi, I would say not even try IX on the DIY. It is designed for specific dimensions of Heat spreaders and also has a high failure rate in application. I think best choice is Coolaboratory Ultra under IHS and also on the IHS. I use this combination to mine with a Noctua NH-D14 and I am very pleased. The nature of Ultra is of a kind, that will never have a pump out effect because is pure metal (alloy that consist of metals that you cannot separate with this kind of pressure an temperatures).All oter TIMs are mixtures of some kind of silicone.So sooner or later degradation and separation of the ingedients will happen. One last thing, if you use the Ultra, smear extremely small quantity in every application (on and under IHS) and be patient while you do it. Over the internet I saw lots of videos and images using huge quantities. The result you want is the best possible contact with the less TIM in between. With my 3770K I achieved 5000 G on air (NOCTUA) and 1.425Volts on Core. During 100% load temps were playing around 74-77 with max spiking to 81 (ambient 20).