• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

What is the bottom layer of the CPU die?

Joined
Feb 22, 2019
Messages
10 (0.00/day)
I am sanding a CPU die, trying to take a die shot of it.
Many people suggest that I should keep sanding until I can only see the bottom copper layer of the die, and it is very reflective.
The thing is, how do I know if it is the last layer? And what will happen if I continue to sand after the last copper layer is exposed?
 
Sand. Take picture(s). Sand a bit more, take more pictures. Keep sanding. Keep taking pictures.
Eventually you'll run out of sand. Or of card space for the pictures.

This is not a troll post... when doing such a non-reversible operation, just take pictures all the way through the process.

You can then choose the best pictures that fit your goal.
 
Might be worth Googling for videos on doing this. Sounds like that would really help with this kind of task.

Share that picture here once you've got it!
 
You mean like this? :D:D:D

On the serious note, the best way is either manual grinding on a glass with polishing paste. It'll take a few hours, but you'll get the result eventually.
Alternatively - cheap chinese grinder + polishing tip + paste. Basically the same approach, but faster and less accurate (you can "erase" parts of circuitry while working on nearby areas).
 
I'm not going to lap "any CPU die" unless someone can prove I can get over 10c improvement. It's just not worth it for such a small gain.
 
I'm not going to lap "any CPU die" unless someone can prove I can get over 10c improvement. Risk is too high for such small gains, just not worth it.
They're not talking about lapping. :slap: It's about sanding the silicon off the die to look at the bare circuit.
 
I think you want to do something like this:


@delshay this is not about lapping to get some temp improvement, this is about taking a sweet photo.
 
I think you want to do something like this:


@delshay this is not about lapping to get some temp improvement, this is about taking a sweet photo.

OPPs sorry my bad.

Yes I should be able to do that on my AOYUE BGA9000A, even thou I have never done this. I just add a copper sheet on top of the Preheater & place CPU. It's the nearest thing to a hotplate which I don't have.
 
I messed around with this myself a little while ago, and I recommend trying to decap an old school bonded wire type IC first because it's much easier to do. You can use heat and sulfuric acid to dissolve the epoxy packaging or potting. If you do it right it'll even still work afterwards.
 
I'm not going to lap "any CPU die" unless someone can prove I can get over 10c improvement. It's just not worth it for such a small gain.
This would have been such a good parody of the extremes some people go to for miniscule cooling improvements LOL
 
Just to make sure this is clear, modern CPUs are in flip-chip packaging. To expose the die you must sand from the BOTTOM of the CPU, not the top.

Also regarding decapping packages using sulfuric acid: battery acid may not be strong enough. You'd need the 98% concentrated variety and may have to heat it as well. Fuming nitric acid is the best acid to use but good luck getting hold of that. Remember that getting either on your skin will immediately cause chemical burns.
 
Last edited:
I am sanding a CPU die, trying to take a die shot of it.
Many people suggest that I should keep sanding until I can only see the bottom copper layer of the die, and it is very reflective.
The thing is, how do I know if it is the last layer? And what will happen if I continue to sand after the last copper layer is exposed?

For the casual reader ... while this is being done for a spe cific purpose, do realize that CPUs and heatsinks are fabricated to have concave / convex matching mating surfaces and lapping, again not the apparent goal here, voids the CPU and some heat sink warrantees.
 
Saw that , oof

EDIT: Where did u start sanding, from the pins or from the IHS?
From the pins side, but remember to remove them (if the CPU has any) as well as any capacitors before sanding.

If you use the rosin method pour out the used rosin and chip fragments before the rosin solidifies. Don't wait for it to cool like I did, but don't pour it directly into the acetone either, as that may result in the acetone vaporizing and blowing molten rosin everywhere! I'm still investigating what to do, but I expect pouring it into cooking oil should cool it off safely (but still be careful).
 
Last edited:
I think you want to do something like this:


@delshay this is not about lapping to get some temp improvement, this is about taking a sweet photo.

I watched this entire vid and forgot I was in this thread, have a like.
 
Some hints to removing the solder:

Adding 60/40 leaded solder will lower the temperature of the solder that holds the pins on; you can just cover the whole thing with solder, and wipe the pins off with a soldering iron.
A 250W weller gun is great for this, and a leather pad will protect your desk while keeping the heat in the chip.

Removing the epoxy:
Once you have the interposer clean, you can dissolve the epoxy that's holding it together with either heat or acid.
Muractic acid or phosphoric acid work pretty well.
Muractic is HCl, and is in driveway cleaner, and phosphoric acid is in mag wheel cleaner.
Don't get either on anything you want to keep, but neither will eat gold or silicon.

Wearing gloves and goggles should be obvious.

We used boiling pyrophosphoric acid to dissolve epoxy; I wouldn't recommend that unless you are a chemist.

Getting the chip off the interface board:
Once you get the epoxy off, it's a BGA chip sitting on a carrier with high temp solder balls, so just unsolder it.
You'll have to go over 500F, most likely.

It's possible you might be able to etch the solderballs off with PCB etching compound made of copper chloride, which is muractic acid and hydrogen peroxide, with some copper dissolved in it; it will eat copper, but not gold or silicon.

You might also be able to get it loose by running a current thru a solution of muractic acid and water; use the chip for the anode, and the etching solution will eat the tin off, and depositing it on the cathode. (I may have those backwards; I'll give it a try tomorrow.)
 
Some hints to removing the solder:

Adding 60/40 leaded solder will lower the temperature of the solder that holds the pins on; you can just cover the whole thing with solder, and wipe the pins off with a soldering iron.
A 250W weller gun is great for this, and a leather pad will protect your desk while keeping the heat in the chip.

Removing the epoxy:
Once you have the interposer clean, you can dissolve the epoxy that's holding it together with either heat or acid.
Muractic acid or phosphoric acid work pretty well.
Muractic is HCl, and is in driveway cleaner, and phosphoric acid is in mag wheel cleaner.
Don't get either on anything you want to keep, but neither will eat gold or silicon.

Wearing gloves and goggles should be obvious.

We used boiling pyrophosphoric acid to dissolve epoxy; I wouldn't recommend that unless you are a chemist.

Getting the chip off the interface board:
Once you get the epoxy off, it's a BGA chip sitting on a carrier with high temp solder balls, so just unsolder it.
You'll have to go over 500F, most likely.

It's possible you might be able to etch the solderballs off with PCB etching compound made of copper chloride, which is muractic acid and hydrogen peroxide, with some copper dissolved in it; it will eat copper, but not gold or silicon.

You might also be able to get it loose by running a current thru a solution of muractic acid and water; use the chip for the anode, and the etching solution will eat the tin off, and depositing it on the cathode. (I may have those backwards; I'll give it a try tomorrow.)
If i dont remove the chip from package (leaving it on the pcb) and sand the cpu from this IHS side of it can i still see the circuits after sanding down the cpu (i know it will take a lot more time) or seeing the circuits from this side is not possible?
 
You'll eventually hit circuitry, but instead of the interconnects, you'll see the tubs.

It will be a lot lower complexity, I guess I'm trying to say.

layout5.png


Check out the above picture for reference.
If you grind it from the top, there's a bunch of detail; if you grind it from the bottom, there's much less.
 
You'll eventually hit circuitry, but instead of the interconnects, you'll see the tubs.

It will be a lot lower complexity, I guess I'm trying to say.

layout5.png


Check out the above picture for reference.
If you grind it from the top, there's a bunch of detail; if you grind it from the bottom, there's much less.
Thanks a lot for the detailed answer much appreciated.

Do you think it would still be interesting to look at with all the shinyness and colors and will look like close enough to the other side? (i want to try this because i dont have heating tools to take the chip out of the package)
 
It will look cooler than just the blank bottom of the die; Art is in the eye of the beholder, however.
Whether it looks cooler/not cool is for you to decide.

Try it; if you don't like it, all you've lost is time.

Scrub on it while you're watching a show, and nothing is lost. :)

You probably won't see the range of colors; part of that is the metallization and interconnects, and those are topside only.
I'm not positive if you will see a difference in the epitaxial Silicon vs. the N+ and P- Silicon.
But you might. :)
 
It's a lot of fun dissecting a processor. Any good enthusiast has done it once or twice I'd like to think..... I know I've done it.

Most questions have been answered, so I could only give my experience.... but meh.

I found out later down the road, you can just purchase a CPU wafer right from e-bay.

 
Back
Top