• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

How do Yields Improve In CPU Production

C

clarkey01

Diamond Member
What are the little tweaks that improve yields ? how do they happen ?. I know there's a few Intel Lithograph workers here that might know some stuff.
 
Originally posted by: clarkey01
What are the little tweaks that improve yields ? how do they happen ?. I know there's a few Intel Lithograph workers here that might know some stuff.
Click to expand...

You mean in the actual lithography system or the silicon layout. I remember when I was designing a chip for my graduate school there were all these design rules to improve yield. The most common and easiest to do was to maintain a mostly uniform metal density to prevent over-etching.
 
One of the guys here just finished his PhD thesis and was recruited by Intel for his work. Basically, there are two things that make CPUs very difficult to produce repeatably: wafer formation and transistor formation.

Wafer formation is carried out under very specific conditions such that the flows are controlled by the solution surface tension (the name of this kind of flows escapes me at the moment). This is the case because, under 'normal' (gravity-driven) flows, a wafer of non-uniform thickness would result.

Transistor formation in the future (not exactly sure how it's done now 😛) will be carried out as a self-assembly process. Given specific conditions (temperature, pressure, concentration of various species), one can predict the formation of crystals very well using mesoscale simulations. I have a thesis on this type of work for anyone who is interested, though it's pretty lengthy (282 pages).

Edit: So, to answer the OP's question, models for these things are still under development. Much of the work done is empirical, so they naturally improve the control of important process variables over time. Every time you shrink the transistor size, you run into a more complicated problem and your empirical work is useless. If you rely on mechanistic models and simulations, however, they are readily extended to different length scales (if they were properly formulated to begin with, of course).
 
I would like to read Your thesis. This kind of stuff fascinates Me.....makes Me wish I took a different field.
 
You were a student at WashU, Cyclo? I live in University City and work in the Loop. Small world 🙂
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top