Width calculation

I'm a relative beginner to extraction routines, and I was wondering if
someone could help out here. I am trying to measure the resistance of
a dog-bone resistor created in an SOI process. The layout of the
device looks like the following: (I apologize for the poor ASCII
diagram)
______
| |
| [] | _ Resistor Head with contact in center
|__ __| /\
| | |
| | |
| | | Resistor Body (Length Parameter)
| | |
__| | __ \/
| | _
| [] | Resistor Head with contact in center
|______|

|<->| Resistor Body (Width parameter)

My problem is the resistor body goes from contact to contact, and is
surrounded by material not factored in to the resistance equation. Is
there a way to directly calculate the width or length of the resistor,
as is stretches inside the head from contact to contact? I can
measure the perimeter, and calculate the width from there (or vice-
versa), but I need either width or length to calculate the other.

 

I assume this is a poly-silicide process.
The resistor body is where the silicide is blocked.
This is what shoulld be calculated.
The R value of the head is so low compared
to the body that it is discarded. Check your manual.
I bet gate poly has a low value & Res-poly has a high.
Colin

 

For generic technologies, the end effects are often left out.

I have seen several analog flows that mandate a more accurate
approach.

I.E.

R = (Rho * Leff / Weff) + 1/Weff * Rend

Where the value of Leff is typically shorter than the distance between
the end
contact array structures by a required/mandated amount. This is
sometimes forced
by the inclusion of End implants to reduce variability in high
precision resistors.

The way that I generate the Rbody is to oversize the end contact
structures
(usually by some amount that is both ( greater than the end contact
spacing)
and ( just! smaller than the Rend to Rend spacing) and then I remove
this from
then Resistor structure and then I oversize the result by the
difference between
this number and the min/expected Rend to Rbody spacing. This becomes
the Rbody
of the resistor and is the device recognition layer.

The remander of the Resistor layer is added to or becomes its own
conductor.

The value of Rend should be supplied by the fab. ( note that in cases
where the
end resistance is << rho, it can be that any Rend effect is swamped by
resistor
tolerance ... and is often ignored.

YMMV