Hello, All,

Currently I have been simulating a multi-layer interdigitated capacitor for 60GHz applications in ADS 2008U1. I've tried every combination of metal and via models, namely, sheet conductor, thick metal expansion both up and down for the metal, and lumped via, 2D/3D distributed for the via.

With the following mesh setups:

- Mesh density: 20 cells/wavelength

- Edge mesh: automatic size

- Mesh reduction

- Horizontal side currents

I've observed the following simulation results:

1. Metal model: sheet conductor

Warning on accuracy (i.e., unphysical behavior).

2. Metal model: thick metal expansion

The comparison between the up or down metal expansion gives close results, w.r.t., the intended and parasitic capacitance. As expected, compared to 2D/3D distributed via model, the lumped via model (zero via thickness) gives less capacitance for a multi-layer interdigitated capacitor. There is little difference between 2D and 3D via models.

To balance between accuracy and simulation speed, I'll probably go for thick metal expansion (either up or down) and 2D via model for the multi-layer interdigitated capacitor simulation. However, I happened to see some comments about MIM cap simulation in Momentum,

" Don't use thick metal expansion to model MIM caps since a hollow thick conductor model can not model properly 90-degree current rotation in MIM cap metal plates."

- from Youri V. Tretiakov, RFMD

RF Passive Device Design, Modeling and Measurements (BCTM 2006 Short Course).

My question is: what's your vote on the metal and via option for an accurate interdigitated capacitor simulation?

Best Regards,

yanyujin

Currently I have been simulating a multi-layer interdigitated capacitor for 60GHz applications in ADS 2008U1. I've tried every combination of metal and via models, namely, sheet conductor, thick metal expansion both up and down for the metal, and lumped via, 2D/3D distributed for the via.

With the following mesh setups:

- Mesh density: 20 cells/wavelength

- Edge mesh: automatic size

- Mesh reduction

- Horizontal side currents

I've observed the following simulation results:

1. Metal model: sheet conductor

Warning on accuracy (i.e., unphysical behavior).

2. Metal model: thick metal expansion

The comparison between the up or down metal expansion gives close results, w.r.t., the intended and parasitic capacitance. As expected, compared to 2D/3D distributed via model, the lumped via model (zero via thickness) gives less capacitance for a multi-layer interdigitated capacitor. There is little difference between 2D and 3D via models.

To balance between accuracy and simulation speed, I'll probably go for thick metal expansion (either up or down) and 2D via model for the multi-layer interdigitated capacitor simulation. However, I happened to see some comments about MIM cap simulation in Momentum,

" Don't use thick metal expansion to model MIM caps since a hollow thick conductor model can not model properly 90-degree current rotation in MIM cap metal plates."

- from Youri V. Tretiakov, RFMD

RF Passive Device Design, Modeling and Measurements (BCTM 2006 Short Course).

My question is: what's your vote on the metal and via option for an accurate interdigitated capacitor simulation?

Best Regards,

yanyujin

Mesh Reduction is another setting that can cause inaccuracies in IC type structures (although that may change in ADS2009 where it has been implemented in a more robust manner). So I recommend to always leave Mesh Reduction off, or at least compare before using.