Interesting article about making SOI wafers

Status
Not open for further replies.
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

johnno wrote:
> "YKhan" <yjkhan@gmail.com> wrote in news:1126025088.180864.244280
> @g47g2000cwa.googlegroups.com:
>
> > http://www.forbes.com/home/free_forbes/2005/0919/072.html
>
> Wow. I can see why Intel don't want to use SOI - $1000/wafer vs $300. That
> is a massive chunk of wafer cost. Or maybe they were comparing 300mm SOI
> wafers with 200mm normal wafers.

Yeah, that cost sort of surprised me too. However, I don't think that's
the cost of a 300mm wafer, that's probably the cost of 200mm wafers. I
had heard somewhere that the cost of a 300mm wafer is about $5000! And
I'd assume that that means non-SOI 300mm. If the cost of converting
natural to SOI wafers is a fixed cost, then a $5000 wafer would go upto
$5700 (i.e. additional $700). If the cost is proportional to the area
of the wafer, then the cost will go up by $1400 -- 300mm wafers have
double the area of 200mm ones. In either case it looks like the
proportionate cost increase of natural vs. SOI is much lower with 300mm
than with 200mm.

Yousuf Khan
 
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

"YKhan" <yjkhan@gmail.com> wrote in
news:1126037557.929669.82300@o13g2000cwo.googlegroups.com:

> johnno wrote:
>> "YKhan" <yjkhan@gmail.com> wrote in news:1126025088.180864.244280
>> @g47g2000cwa.googlegroups.com:
>>
>> > http://www.forbes.com/home/free_forbes/2005/0919/072.html
>>
>> Wow. I can see why Intel don't want to use SOI - $1000/wafer vs $300.
>> That is a massive chunk of wafer cost. Or maybe they were comparing
>> 300mm SOI wafers with 200mm normal wafers.
>
> Yeah, that cost sort of surprised me too. However, I don't think
> that's the cost of a 300mm wafer, that's probably the cost of 200mm
> wafers. I had heard somewhere that the cost of a 300mm wafer is about
> $5000! And I'd assume that that means non-SOI 300mm. If the cost of
> converting natural to SOI wafers is a fixed cost, then a $5000 wafer
> would go upto $5700 (i.e. additional $700). If the cost is
> proportional to the area of the wafer, then the cost will go up by
> $1400 -- 300mm wafers have double the area of 200mm ones. In either
> case it looks like the proportionate cost increase of natural vs. SOI
> is much lower with 300mm than with 200mm.
>

$5000 for a processed 300mm 90nm wafer sounds about right to me, maybe a
bit high, but I'm only familiar with 200mm, and .18um at that.
But we're still looking at either 14% or 28% increase in wafer cost -
either way it makes a big dent in your margins.
 
Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

johnno <johnno@nospam.invalid> wrote:
> "YKhan" <yjkhan@gmail.com> wrote in
>> Yeah, that cost sort of surprised me too. However, I don't
>> think that's the cost of a 300mm wafer, that's probably the
>> cost of 200mm wafers. I had heard somewhere that the cost
>> of a 300mm wafer is about $5000! And I'd assume that that
>> means non-SOI 300mm. If the cost of converting natural to
>> SOI wafers is a fixed cost, then a $5000 wafer would go upto
>> $5700 (i.e. additional $700). If the cost is proportional
>> to the area of the wafer, then the cost will go up by $1400
>> -- 300mm wafers have double the area of 200mm ones. In
>> either case it looks like the proportionate cost increase
>> of natural vs. SOI is much lower with 300mm than with 200mm.

> $5000 for a processed 300mm 90nm wafer sounds about right to me,
> maybe a bit high, but I'm only familiar with 200mm, and .18um
> at that. But we're still looking at either 14% or 28% increase
> in wafer cost - either way it makes a big dent in your margins.

Some perspective? A 300mm wafer should yield about 500 dice
(100mm2 ea, incl kerf & layout losses). So that's $10/ea, multiplied
by some yield factor. 28% is $2.80/CPU -- say 3% on $100 ASP.

I wouldn't call that a big dent unless your yields are < 60%

-- Robert
 
Status
Not open for further replies.