question about contact breakdown

[020170]

I have a question about "contact breakdown".

Is there a possibility that contact layer between active and met1 experience breakdown process while operating with high voltage mosfet?

It is said to use long contact layer to prevent contact break down.

However I don't know why dons't not happen contact breakdown when I use long contact layer.

1. What is the meaning on "contact breakdown?"

2. Why does't happen contact breakdown while using large and long contact layer?

3. In high voltage mosfet, how can occur contact breakdown sequentially?


I'll appreciate your help to explain in the view of layout engineer level.

thanks.

 

[deepak242003]

I believe contact breakdown is possible if your contact is not able to handle the amount of current flowing through it... as a solution either you can increase the contact size or number of contact.. where later is more prefered as it decreased your contact resistance..

let me know if there is any other view.

 

[dick_freebird]

It's not clear to me from the description, what this means.

If the drain metallization is too close to the gate it might
result in some field enhancement over the drift region,
reducing breakdown voltage and frying the FET. Or maybe
the "second gate" oxide itself can be punched through
by drain voltage if the metal gets too close. Perhaps
the way
they stack regions, long contact pushes that back.

"long contact layer" could also be providing some ballasting
effect, preventing one contact from "hogging" current
and failing early.

We'd have to see some pictures to understand the meaning.

 

[kapilsn]

I thought that the size of the contact is predifined and cant be increased or decreased.
The only thing we can do for reliability of the circuit to put more contacts.

 

[dick_freebird]

I used to work in older technologies where only the minimum
contact dimension was specified. However most modern ones
enforce a single, uniform dimension.

It might be however, that the drain "contact" includes
the N+ region as well, like a PCell or a stretch feature thereof,
and the naming is just sloppy.

Like ESD cells where you have an extended S/D contact
region for ballasting, the contact itself only moves but the
N+ stretches.