Hello Hagen, You are right, this is a four-terminal (a.k.a. Kelvin-contact) approach with one pair of conductors for voltage sense and another for current forcing. The metal1-ndiff is a bit different from the others, but this is a compromise: I had to squeeze it between the two pads, and I did not want the perpendicular diff. track to be meandered back under the pad. The reason for the well contact is to ensure that the well is biased to a known potential. Normally, it would be a fifth pad used as a kind of "shield", but since the equipment has only four probes, I chose to tie it to one of the current force leads. The reason for not having a separate pdiff (or other front-end layer) contact is that in the process all contacts will have a silicide layer over the diffusion, and that will cause the semiconductor to be degenerated (neither p, nor n-doped, with the interface between the silicide and the underlying doped layer is either ohmic or degenerate schottky barrier), so contact resistance will be mainly determined by the silicide layer, that is assumed to be the same for all front-end layers. (the behavior of the silicide-doped layer interface is handled in the MOSFET characterization.) Regards, Ferenc
On Tue, Nov 6, 2018 at 5:46 AM Hagen SANKOWSKI hsank@posteo.de wrote:
Hello Ferenc.
I like to re-add your measurement structures for contacts again into the Pearlriver_quarter structure.. unfortunately I have to re-arrange the test pads..
Well, looking at your structures I am wondering, what your design intentions are. I assume you´re like to do "Kelvin-Structures", e.g. described at [0]. This are implemented 'clean' for via and via2. In case of poly and ndiff i did not fully get the points.
Can you explain me (or the list) in more detail? E.g.
- Why you connect the pwell also/only on one side?
- Why this are not completely symmetric on all axis?
- Why there is no pdiff contact structure also?
Just for curiosity also, I always like to learn :-)
Regards, Hagen.