Intel Production Process

[Fungalberry]

You may/may not find this interesting:

I just went to a local computer store (holy sh1t those guys are smart!) and they told me about how Intel makes their Pentium 4 and Celeron Processors. OK, so they make a whole ton of Pentium 4 CPUs. They are arranged on shelves in a big warehouse, about 6x12 for each shelf thingy. The ones on opposite ends of the warehouse tend to be more defective, so these are taken off the racks and many of the features are disabled on these processors. These become Celeron Processors. The ones in the center, which tend to be less defective, are kept as they are and sold as Pentium 4s. :? Strange. I'm not making this up, it's fact.

 

[Grimmy]

Well, if companies want to be efficient in selling... thats what they do.

I knew a friend that used to work at a slicon manufactoring company. Back in the days when celeron came out, intel would by the lowend left over silicon to make celerons.

That was one reason why I stayed away from anything labeled celeron, since I knew they were using the cheapest material/method of making them.

I would only assume AMD doing the same things with the durons, when they came out.

 

[Fungalberry]

What, you are shocked??? AMD does exactly the same thing....

Yeah, it's actually kinda surprising. I wasn't posting that to beat up on Intel, I was just posting it because I thought it was interesting. It's not that surprising that AMD does it too.

 

[7H4_D00D3]

1. A wafer is created from sand. Very long process, but it happens.

Probably you already know that, but I'll still post it here
The chemical process goes like this:

Si+C=---->>>Si2O+CO2

Si2O+Cl----->2Si+ClO (not sure about this one)--->>>95% pure Si, not enough for electronic devices

And after another process I gets 97% pure Si, just enough for electronics

 

[Grimmy]

I did say 'back in the day, when celerons came out'.

My friend explained to me, at 1st they didn't know why intel wanted to buy the 'crappy/lowend' silicon waffers that they produced, until, the celeron went out.

So I'm talking on the past. I don't have the knowledge of what they do today, which may be more current in what you describe in your post.

So, sorry if I confused ya.

 

[7H4_D00D3]

1. A wafer is created from sand. Very long process, but it happens.

Probably you already know that, but I'll still post it here
The chemical process goes like this:

Si+C=---->>>Si2O+CO2

Si2O+Cl----->2Si+ClO (not sure about this one)--->>>95% pure Si, not enough for electronic devices

And after another process I gets 97% pure Si, just enough for electronics

Yeah, you got me goin', your first reaction is partially correct.... second reaction is not. (side note, the purest sand comes from the deserts of Saudi Arabia not only do they have us on oil, but they have us on sand too )....

SiO2 + C --> Si + CO2 (Si is not very pure).

Si + 3HCL --> SiHCl3 (Gas) + H2(gas, which is removed).

2SiHCl3 + H2 (High pressure) -- > 2Si + 6HCl (Gas) (Si is pure).

Single crystals are then grown by a method called CZ (Czochralski method). Essentially, you heat the silicon up to 1400 deg C, the melting point of silicon, form molten Si. Using a seed crystal literally tied to the end of a wire, you dip it in the molten Si briefly and begin pulling upward. As you pull Si adheres to the seed crystal and itself crystalizes. The purity levels in CZ silicon are typcially acceptable, but you can further purify Si by another method called Float Zone (FZ), this one is fascinating but too technical to go into details. When all is said and done, impurity levels in electronic grade silicon drop to the few part per million regime (PPM), typcially consisting of Fe, Al, Ca, Mg, Na, Zn.

The science of pulling a single crystal Si boule of high enough quality for 200 mm wafers is tough enough, 300 mm runs into a multitude of problems, the pull rate, the thermal cool rate, center to edge crystalization defects.... etc.

I have used this reference many times, it is my go-to text for semiconductor manufacturing basics: Solid State Electronic Devices (B. Streetmen, S. Banerjee)

Very interesting stuff....

Sugoi !!!

 

[uber_g]

Well, if companies want to be efficient in selling... thats what they do.

I knew a friend that used to work at a slicon manufactoring company. Back in the days when celeron came out, intel would by the lowend left over silicon to make celerons.

That was one reason why I stayed away from anything labeled celeron, since I knew they were using the cheapest material/method of making them.

I would only assume AMD doing the same things with the durons, when they came out.

-true