what W/S mean in metal resistance model?

[flushrat]

library tt_res

in 80nm hspice library
there are three the metal 1 resistance model: rm1l rm1s rm1w
rm1l: metal 1 with W/S = 0.108/0.324
rm1s: metal 1 with W/S = 0.108/0.108
rm1w: metal 1 with W/S = 0.324/0.162

what does W/S mean?

 

[dipswitch]

metal resistance w/s

Typically, when referring to metal, W/S means Width / Space. I.e., the different metal pitches one can use in layout.

The confusing thing here is that the space should not affect resistance, so it would not be included in a resistance model. However, it will greatly effect the capacitance, and therefore would be included in any model that has capacitance.

If you post the models, we could tell you for sure.

 

[jcpu]

w/s mean

The confusing thing here is that the space should not affect resistance

Sometimes space does affect resistance.
while photo mask is the same, as well as the etching recipe.
The etching rate has something to do with the circumstances,
which they generally call "density".
For example, whether if it is stand-alone metal or with dummy traces besides is very different.

Of course, we are not sure if this is exactly the purpose of foundry modeling engineer until we can see the model itself.

 

[flushrat]

here is part of the model, the only difference between metal resistance models is the sheet resistance.
.LIB TT_RES
.Param
......
r_rm1l = 0.0791 r_rm1s = 0.115 r_rm1w = 0.109
.......
.ENDL TT_RES
*********************************** Metal 1 with W/S=0.108/0.324 ***********************************
.subckt rm1l n1 n2 l=length w=width scale=1 m=1
.param Rsh = r_rm1l
.param dw = 0
.param pvc1 = 0 pvc2 = 0 ptc1 = 2.846E-03 ptc2 = -5.351e-07
.param pt='temper'
.param tfac='1.0+ptc1*(pt-25.0)+ptc2*(pt-25.0)*(pt-25.0)'
r1 n1 n2 'rsh/m*l*scale/(w*scale-dw)*(1+pvc1*(abs(v(n2,n1))/l/scale)+pvc2*(v(n2,n1)/l/scale)*(v(n2,n1)/l/scale))*tfac'
.ends rm1l

*********************************** Metal 1 with W/S=0.108/0.108 ***********************************
.subckt rm1s n1 n2 l=length w=width scale=1 m=1
.param Rsh = r_rm1s
.param dw = 0
.param pvc1 = 0 pvc2 = 0 ptc1 = 2.846E-03 ptc2 = -5.351e-07
.param pt='temper'
.param tfac='1.0+ptc1*(pt-25.0)+ptc2*(pt-25.0)*(pt-25.0)'
r1 n1 n2 'rsh/m*l*scale/(w*scale-dw)*(1+pvc1*(abs(v(n2,n1))/l/scale)+pvc2*(v(n2,n1)/l/scale)*(v(n2,n1)/l/scale))*tfac'
.ends rm1s

*********************************** Metal 1 with W/S=0.324/0.162 ***********************************
.subckt rm1w n1 n2 l=length w=width scale=1 m=1
.param Rsh = r_rm1w
.param dw = 0
.param pvc1 = 0 pvc2 = 0 ptc1 = 2.846E-03 ptc2 = -5.351e-07
.param pt='temper'
.param tfac='1.0+ptc1*(pt-25.0)+ptc2*(pt-25.0)*(pt-25.0)'
r1 n1 n2 'rsh/m*l*scale/(w*scale-dw)*(1+pvc1*(abs(v(n2,n1))/l/scale)+pvc2*(v(n2,n1)/l/scale)*(v(n2,n1)/l/scale))*tfac'
.ends rm1w

 

[dipswitch]

Actually, jcpu is correct, microloading effects of the plasma metal etch will affect the actual width versus the drawn width. However, two things: first of all, we place dummy metal fill to make the density as uniform as possible. Second, I have never seen a 40% increase in rsh due to this effect. It should be significantly smaller, especially between the 0.162 space and the 0.324 space, because this should be easily within the process spec at 80nm.

I can't come up with a definitive answer. Anyone else?