waffle transistor structure

[deepak242003]

Any body has proven experience wiht waffle transistor structure..?? ..
I am really concern about sharp corners in channel.. Is it really helpful against conventional transistor
??

 

[dick_freebird]

We used to use waffle layouts in analog switches. The thinking
was that it would produce the highest width per area and lowest
on-resistance. But that was back when metal was fat and
gate oxides were thick, so FET on-resistance dominated.

Going through layout trade studies however, the benefit
is small and there are drawbacks like increased capacitance,
and the field concentration is certain to elevate leakage
and likely to detract from reliability.

CAD systems may do a pretty poor job of extracting the real
effective device width. The "corner square" is longer channel
than the rest of the FET and you can't get uniform conduction.
The hexagonal scheme packs better / more uniformly. But
there, you may run afoul of non-45 /non-ortho geometry rules
if you're in an "advanced" foundry, and square contacts
probably push slightly worse minimum nesting.


All in all I've come to consider it a waste of effort and don't
bother with it anymore. Even in powerMOSFET world, the
hexfet has pretty much been squashed by optimized stripe
geometries.

 

[deepak242003]

many thanks for your view.. I fully agree with you...

If you allow to ask....
What is your view regarding serpentine MOSFETs... for long transistor.... I have seen few of them in some proven Power pads(0.13um)... However I am corcened wiht sharp corners in channel....

Please share you view s..


Deepak.

 

[llbaobao]

I do not know much about this.

But I do see the waffle transistor structure in some tape out designs(not powerMOSFET), and the result is OK.