A few questions regarding wafers

[blyndy]

Hi, there are often photos of people holding or otherwise touching an exposed wafer. http://regmedia.co.uk/2007/09/19/idf_wafer_1.jpg

Aren't they dust sensitive? are they 1.used/2.usable after being displayed or handled?

What about this image: http://eetimes.com/galleries/slideShow.jhtml?galleryID=74&imageID=9, used and/or usable?

Isn't an individual CPU wafer worth tens if not hundreds of thousands of dollars and take months to produce? If they do 'sacrifice' some wafers for testing, is it closer to 1 in 100 or 1 in 1000?

P.S. Is everyone having to reset their passwords?

 

[KIAman]

Other than the initial R&D for producing the chip design and fitting the fab for design production, the cost of the wafer is just the cost of the wafer in raw form. That's why downsizing chips to a smaller process equate to cheaper chips.

 

[bobsmith1492]

Other than the initial R&D for producing the chip design and fitting the fab for design production, the cost of the wafer is just the cost of the wafer in raw form. That's why downsizing chips to a smaller process equate to cheaper chips.

So wafer pics where they are being handled are probably just unprocessed wafers?

 

[William Gaatjes]

P.S. Is everyone having to reset their passwords?

I had to reset my password too.

And since the new forum style, while logged in i have sometimes to log in again when i want to post a message after pressing the [submit reply] button. I get the following message while logged in : "You are not logged in"
It is not every time and it seems like some random time out bug/feature.

 

[TuxDave]

Hi, there are often photos of people holding or otherwise touching an exposed wafer. http://regmedia.co.uk/2007/09/19/idf_wafer_1.jpg

Aren't they dust sensitive? are they 1.used/2.usable after being displayed or handled?

What about this image: http://eetimes.com/galleries/slideShow.jhtml?galleryID=74&imageID=9, used and/or usable?

Isn't an individual CPU wafer worth tens if not hundreds of thousands of dollars and take months to produce? If they do 'sacrifice' some wafers for testing, is it closer to 1 in 100 or 1 in 1000?

P.S. Is everyone having to reset their passwords?

It's a business of volume. A large amount of money goes into the litho machines that process the wafers so if you divide up the total cost by the # of wafers you produce, you better be making a ton of wafers to make it worth your time or else the wafers will cost hundreds of thousands of dollars.

The incremental cost of a wafer isn't that large (relatively speaking) and so churning one out that looks pretty for display purposes isn't outrageous.

 

[KIAman]

For comparison, $48 will buy you a 300mm wafer retail http://cgi.ebay.com/ws/eBayISAPI.dl...f98c390&itemid=260471216139&ff4=263602_263622

I'm sure the chip manufacturers get a deeper discount. The point is that past the cost of R&D and refits of the fabs, the cost to produce a finished (etched) wafer is almost nil.

If they destroy 10 finished wafers because of Marketing and PR, then they will simply throw 10 more wafers through the printing (lithographic) machines at the cost of 10 bare wafers + change (electricity).

This is the reason for cheaper processors as they get smaller (more chips per wafer) and also speed binning = profit (selling lesser performing chips is better than not selling them at all).

 

[blyndy]

And since the new forum style, while logged in i have sometimes to log in again when i want to post a message after pressing the [submit reply] button. I get the following message while logged in : "You are not logged in"
It is not every time and it seems like some random time out bug/feature.

That's happening for me too.

 

[silverpig]

They are very dust sensitive and generally cannot be used if taken out of the clean room environment unless they undergo a vigorous cleaning procedure again.

<--- worked in a nanofab

 

[Net]

each one of those squares is called a die.

there will be imperfections spectacled randomly around the wafer. each die that has a imperfection on it is tossed out.

The smaller the die the less the change of having an imperfection on a given die and the less amount of die that need to be tossed out per area of wafer.

all die that aren't a full square are tossed out too. so parts of the edge of the circle contains unusable die.

yes, a dust particle can make a large difference in conductivity. they won't be selling that wafer, they'll keep it around as their show wafer.

the actual cost of the material per die is in the cents range. Running multimillion dollar machines in a clean room with expensive employee's and years of R&D ups the cost. i saw an article that said $40 per die for intel. the article was written 2005.

The machine picks out a sample of wafers from the wafers being produced and test a few of the die, the tested dies are marked with a red dot. imagine that wafer shown with 3 or 5 die that have a red dot, if it was the wafer that was selected for testing.

 

[CanOWorms]

Usually there is some sort of passivation layer so that dust/debris won't be too much of a factor in a finished product.

However, handling a wafer or a single device like in the pictures would lead to it not being sold nor used in testing because that is completely out of ESD requirements.

 

[drinkmorejava]

Other than the initial R&D for producing the chip design and fitting the fab for design production, the cost of the wafer is just the cost of the wafer in raw form. That's why downsizing chips to a smaller process equate to cheaper chips.

That makes no sense. You said wafers ~free but fitting more chips of a free wafer makes them cheaper. The cost has everything to do with limited billion dollar machinery-> limited number of wafers processed and almost nothing to do with wafer cost.

 

[FrankSchwab]

My wife used to work in a 12", 0.25 micron fab. They figured finished wafers cost about $1000 each; that, of course, is not what they charged for them....

We're currently building products on a 0.13 micron process, on 8" wafers, and getting similar pricing, though we're only buying dozens of wafers a year. Once (if!) we ramp up, we should be buying tens of thousands of wafers a year, and getting a much better price.

/frank

 

[djnsmith7]

My wife used to work in a 12", 0.25 micron fab. They figured finished wafers cost about $1000 each; that, of course, is not what they charged for them....

We're currently building products on a 0.13 micron process, on 8" wafers, and getting similar pricing, though we're only buying dozens of wafers a year. Once (if!) we ramp up, we should be buying tens of thousands of wafers a year, and getting a much better price.

/frank

To the OP, yes, any type of wafer is very sensitive to particles, regardless of size or count. Particles are measured in microns & depending on the application, have a maximum particle requirement per wafer, which can be measured in a few different ways, using machinery. If you see someone handling a wafer outside of the cleanroom, that particular wafer is either being used to showcase something or the application isn't particle sensitive, as others have mentioned above.

Frank, if our companies aren't currently doing business together, maybe we should be. Send me a PM next time you need a quote for wafers, regardless of size, & we'll go from there.