Help! Getting to the wafer of a CPU

BlueToast

Junior Member
Feb 13, 2011
2
0
0
www.hlrse.net
I am seeking the assistance of a moderator to help me post this thread.

EDIT: I can't figure out who the moderator of this forum is. If you would be kind, please replace the entire contents of this Post#1 with the raw contents at http://privatepaste.com/bfd258c85a -- I did my absolute best to filter it out of 'promotional links' and names, and if any rule breaking conflicts are found in it, I will fix them to conform to the rules here.
 
Last edited:

BlueToast

Junior Member
Feb 13, 2011
2
0
0
www.hlrse.net
I am having some trouble "deprocessing" CPU dies to get to the wafer. I am trying to attain samples of wafers that I may put under a digital microscope and digital projector for a course lecture at my college that I will have roughly next week (or the week after). I have received many generous CPU donations for this, and have also bought glass etching cream -- which does not seem to work as well as I was hoping (as picture below).






In short, I am seeking for suggestions, advice, and help in my quest for CPU wafers and getting to them.
 

PottedMeat

Lifer
Apr 17, 2002
12,363
475
126
what you want on that cpu is to desolder the die and photograph that. i really don't know how you can get it off in one piece, maybe a heat gun, but the filler material may make that very difficult.

do you have an old 486 like this:
2h39bab.jpg
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
Old thread:
http://forums.anandtech.com/showthread.php?t=2024893#post28930136

REALLY old thread, I don't recommend repeating what I did there:
http://forums.anandtech.com/showthread.php?t=180048


I think you have to attack the die from the other side. If you consider a cross section of a die, there will be hundreds of microns of substrate (uninteresting silicon that's just there for mechanical strength), <1ish micron of transistors (I'd expect you need >50X magnification to see really interesting things here), and 10ish microns of wires (looks cool at medium magnification). You could sand / etch from the top side until you have almost nothing left and still see nothing interesting. If you start from the other side of the die, you're starting at something interesting and it gets progressively more interesting until you reach obvious stopping points (when you get to the smallest features you want to see). Coming from the other side of the die you're also going to be looking "top down", so the metal layers will actually be on top of the transistors.

I recommend just using sandpaper instead of glass etching cream. I recommend trying to use a 65nm or older CPU... 45nm was very difficult to sand evenly.
 

Modelworks

Lifer
Feb 22, 2007
16,240
7
76
I really wouldn't advise doing that as the chemicals involved are seriously toxic as others have said.
Do these have to be specific x86 cpus ?
If not you can find cpu wafers for sale online that might be a better way of approaching this.

Some of the cheaper ones:
http://www.goldmine-elec-products.com/prodinfo.asp?number=G14259
http://www.goldmine-elec-products.com/prodinfo.asp?number=G13710


Also check ebay for cpu die, a lot of times older cpu will be posted there and they are much easier to view than newer ones.

Just looked on ebay and pentium pros are the ones to get, but looks like everyone thinks they are worth their weight in gold.
 
Last edited:

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,077
3,577
126
Id check ebay for keychains or paper weights.

I both intel and AMD have naked dies in those which u can take pictures of very easily.

I recommend just using sandpaper instead of glass etching cream. I recommend trying to use a 65nm or older CPU... 45nm was very difficult to sand evenly.

OMG i havent seen you in ages.. nice to see you back.
 

pm

Elite Member Mobile Devices
Jan 25, 2000
7,419
22
81
First, it's me being pedantic, but the term that you want is the CPU "die" - a wafer is a big disk with lots of CPU's on it.
die: http://en.wikipedia.org/wiki/Die_(integrated_circuit)
wafer: http://en.wikipedia.org/wiki/Wafer_(electronics)

As Ctho9305, bear in mind that the first chip is a BGA device - the parts of the chip you want to photograph are on the other side. So glass etching cream won't work unless you remove the chip and do the other side. You could theoretically come in from the backside (the top side that you can see without removing it from the package) but you've got a lot of grinding/etching to do it before you can get to the metal - which is the pretty part.

I'd follow CTho9305's advice - he's done this before. When I want to take a die photo, I ask for raw unpackaged die. :)