ARM based Opterons incoming!

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Abwx

Diamond Member
Apr 2, 2011
8,792
122
126
^^ This! Thanks! I've been trying to remember where I had seen that graph before.

That graph is misleading.

What doesnt appear is that the dust particles number increase
exponentialy with decreasing particles sizes , hence the smaller
the node the higher the number of particles likely to cause
a defect of the chip , in short what is a microscopic element
at a given transistors size become a rock at smaller nodes.
 

Ajay

Diamond Member
Jan 8, 2001
4,943
114
136
That graph is misleading.

What doesnt appear is that the dust particles number increase
exponentialy with decreasing particles sizes , hence the smaller
the node the higher the number of particles likely to cause
a defect of the chip , in short what is a microscopic element
at a given transistors size become a rock at smaller nodes.
Why increase? For a given clean room rating the number of airborne particles per m^3 should be the same.
 

Abwx

Diamond Member
Apr 2, 2011
8,792
122
126
Why increase? For a given clean room rating the number of airborne particles per m^3 should be the same.
The total weight of a given size range of particles decrease with
size but the number of individual particles increase.

A clean room is specified with different particles sizes per m3 ,
the smaller the particles the more the tolerance of particles/m3
is increased otherwise there would be no clean room, moreover
given that lots of steps in the manufacturing process are done
with an atmosphere wich despite being controled in its composition
and temperature will inevitably have sustained dust.
 

Abwx

Diamond Member
Apr 2, 2011
8,792
122
126
Dont know but you could do a thread about it to check
if this is a purely consumerism orientated forum......
 
Oct 8, 2002
10,577
0
71
That graph is misleading.

What doesnt appear is that the dust particles number increase
exponentialy with decreasing particles sizes , hence the smaller
the node the higher the number of particles likely to cause
a defect of the chip , in short what is a monstrously sized boulder
at a given transistors size become an even bigger boulder
Fixed that for you. ;)
 
Feb 2, 2009
12,873
164
126
Another thing is that on a bigger die, one dust particle could make a small part of the die inactive, allowing you to die harvest the chip. With a smaller die, the same dust particle could make the entire chip non operational completely ruin it.
 

Ajay

Diamond Member
Jan 8, 2001
4,943
114
136
The total weight of a given size range of particles decrease with
size but the number of individual particles increase.

A clean room is specified with different particles sizes per m3 ,
the smaller the particles the more the tolerance of particles/m3
is increased otherwise there would be no clean room, moreover
given that lots of steps in the manufacturing process are done
with an atmosphere which despite being controlled in its composition
and temperature will inevitably have sustained dust.
Good point. I hadn't really thought it through all the way and, as TuxDave added, even the very fine particles are large compared to nano meter sized xtor features.
 
Dec 28, 2013
32
0
41
The total weight of a given size range of particles decrease with
size but the number of individual particles increase.

A clean room is specified with different particles sizes per m3 ,
the smaller the particles the more the tolerance of particles/m3
is increased otherwise there would be no clean room, moreover
given that lots of steps in the manufacturing process are done
with an atmosphere wich despite being controled in its composition
and temperature will inevitably have sustained dust.
Realistically, environmental contamination is almost never a cause of significant yield loss. In a modern production line with FOUPs and mini-environments you can pretty much leave a wafer sitting around in atmosphere for minutes without a significant increase in yield killing particles. (If you can't, your tool is out of spec.)

What will/may get better as a process matures is process induced yield loss, whether due to contamination or uniformity problems. But the start of mass production for a given node always coincides with some satisfactory yield level on a test vehicle, which in turn corresponds to an appropriately low defect density at the appropriate threshold. Yes, this threshold gets smaller for newer nodes, but the processes that induce defects have to be significantly better than the ones used in the older node by the time the new node goes into production.
 
Sep 4, 2007
190
0
0
The total weight of a given size range of particles decrease with
size but the number of individual particles increase.

A clean room is specified with different particles sizes per m3 ,
the smaller the particles the more the tolerance of particles/m3
is increased otherwise there would be no clean room, moreover
given that lots of steps in the manufacturing process are done
with an atmosphere wich despite being controled in its composition
and temperature will inevitably have sustained dust.
That is practically irrelevant to modern day fabs. The cleanliness of the fab air has virtually no impact on wafer defects because the wafers are never exposed to the fab air in 300 mm fabs. The wafers are transported in sealed containers and are only opened to interface directly with the inside of the process tools. Nearly all defects are from the process equipment, whether it be movement related or process related.
 

Abwx

Diamond Member
Apr 2, 2011
8,792
122
126
That is practically irrelevant to modern day fabs. The cleanliness of the fab air has virtually no impact on wafer defects because the wafers are never exposed to the fab air in 300 mm fabs. The wafers are transported in sealed containers and are only opened to interface directly with the inside of the process tools. Nearly all defects are from the process equipment, whether it be movement related or process related.
I know that the waffers are not in physical contact with the manufacture atmosphere but keep in mind that there,s inherently dust inside your sealed boxes as well a inside the process tools , it s not by chance that the waffers
are washed numerous times during their processing...
 
Dec 28, 2013
32
0
41
I know that the waffers are not in physical contact with the manufacture atmosphere but keep in mind that there,s inherently dust inside your sealed boxes as well a inside the process tools , it s not by chance that the waffers
are washed numerous times during their processing...
There really isn't dust that is of consequence inside a FOUP. You can leave wafers in these sealed boxes for a month without defect adders. And wet cleans are used to remove residues from other processes, not environmental contamination.
 
Sep 4, 2007
190
0
0
I know that the waffers are not in physical contact with the manufacture atmosphere but keep in mind that there,s inherently dust inside your sealed boxes as well a inside the process tools , it s not by chance that the waffers
are washed numerous times during their processing...
It really comes down to what your definition of "dust" is. In my viewpoint, I consider dust to be particles floating around in the air and sometimes landing on a surface. This would not be a good description of the environment inside a process tool or the sealed boxes. The atmospheric environment inside these tools is incredibly clean and dust defined in this way would barely show up as a blip on the Pareto.

The most common sources of particles usually comes from parts rubbing together or chunks of process materials flaking off. If you were to define this as dust then you might be somewhat correct. However, wafers are not washed to remove dust or any other type of particles. Wafer are washed to remove process chemicals and such.
 

krumme

Diamond Member
Oct 9, 2009
5,691
59
136
Perhaps they can use my iroomba?
 


ASK THE COMMUNITY

TRENDING THREADS