• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Someone explain CPU Waffer

bobrownik

Gawd
Joined
Jan 15, 2006
Messages
800
ive been wondering for a while why the cpus are made on the round wafers,

this seems like an awful waste of material in comparison with a square "wafer".


here is the picture of what i mean ,




besides that , id like to know what exactly is the wafer made of and how the cpus are made on the wafer , any links to the info appreciated ,
 
By the process you linked to, why dont they cut the wafer in a square before turning it into a cpu, and melt the wasted silicon and re-use it in the process? Wouldnt this make yeilds almost 100% not including defective cpu's?
 
TruthInRuin said:
By the process you linked to, why dont they cut the wafer in a square before turning it into a cpu, and melt the wasted silicon and re-use it in the process? Wouldnt this make yeilds almost 100% not including defective cpu's?
because it doesn't matter. The dice are cut from the wafer and the surplus can be reused. In fact, the circular shape has some big advantages, since during fabrication various chemicals are applied in very thin and even films by spinning the wafer. If it were non circular, it'd likely be more susceptible to vibrations at high rotating speeds.

If you ever end up at the "Deutsches Museum" in Munich, Germany, they have a rather large exhibit about semiconductor manufacturing, in particular on growing Si crystals.

Wiki has a lot of information too:
http://en.wikipedia.org/wiki/Category:Semiconductor_device_fabrication

The basic steps are these:
  • grow a large Si crystal
  • cut said crystal into wafers
  • depending on the process dope the whole wafer, deposit insulator layer or something esle
  • applym photoresist
  • take "picture" of mask over wafer
  • remove non-exposed resist
  • do something, such as: dope, apply conductor layer, etch away layer etc.
  • repeat the previous couple of steps a couple of times
  • cut wafer into dice
  • package/ test dice

there is an AWEFUL lot of "cleaning" steps that I left out, but they really don't matter on a grand scale.
 
TruthInRuin said:
By the process you linked to, why dont they cut the wafer in a square before turning it into a cpu, and melt the wasted silicon and re-use it in the process? Wouldnt this make yeilds almost 100% not including defective cpu's?
Most sand on the world is made of Silicon. It is literally dirt cheap. With silicon it is not the semiconductor material, that makes the chips expensive, it is the energy, chemicals and water used during the process, as well as all recouping R&D and other sunk costs.

Sidenote: yield is defined as the ratio of good dice to all dice. So reusing the silicon, if it even made sense, due to all the junk it/ on it at that time, would not improve yields.
 
Gotcha.

Also, does all dice include the dice that are along the outside of the wafer that cant be used?
 
Typically there is nothing along the outside edges. They just fit as many cores as they can in the available space. For making them think of applying a thin layer of metal to the entire surface, then shooting a laser through a photonegative. That burns off any unwanted metal. Rinse and repeat. Sort of like making a loaf of bread and starting out with slices.
 
I used to work for HP in Ft. Collins Colorado and while I never made it into the clean room, one of my jobs was carrying the round wafer samples up to the QC Desk.
Some were as big as a large soup can, like the Chunky Soups size while some were huge, as big as a number 10 can.

Those things are very fragile and paper thin and I actually broke a few ever now and then.
I remember a few guys would take them home and sandwich them between two pieces of glass to make coke mirrors out of them.
Seems like everyone at HP back then was doing coke
I just remember being amazed at how fragile those things were.
I used to have a stack of them in my junk box, but they eventually all got broken over the years.
I wished I would have made something out of them like a coaster set or something as they were very cool looking.
 
TruthInRuin said:
Gotcha.

Also, does all dice include the dice that are along the outside of the wafer that cant be used?
It is unlikely that someone would include dice that are incomplete in their yield calculations. After all, as a fab, you'd like to boast with high yield numbers.

Anarchist4000 said:
Typically there is nothing along the outside edges. They just fit as many cores as they can in the available space. For making them think of applying a thin layer of metal to the entire surface, then shooting a laser through a photonegative. That burns off any unwanted metal. Rinse and repeat. Sort of like making a loaf of bread and starting out with slices.
I was under the impression that the "laser" was used to expose the resist, while the removal of unexposed resist and underlying material is done chemically, not optically/ thermally.
 
drizzt81 said:
I was under the impression that the "laser" was used to expose the resist, while the removal of unexposed resist and underlying material is done chemically, not optically/ thermally.

That is correct. It is still a lithographic process. The poster you quoted was giving an analogy, I think.
 
Back
Top