MMWave Layout/EM Simulation Questions

[stark21]

Hello Dr Volker,

Thank you for your descriptive post. I am simulating RFCMIM capacitors in IHP PDK.

Do you recommend Momentum for simulating MIM capacitors at high frequencies? If yes, what should the meshing be if we want to accurately simulate the MIM caps at frequencies as high as 200 GHz? The design guide recommends meshing to be greater than 600cells/wavelength for accurate simulation and this setting obviously takes a lot of time. Are there any ways to work around this problem without compromising the accuracy?

Also, how accurate are the capacitor models for IHP SG13S?

Thanking you in advance.

 

[volker@muehlhaus]

Hello noname,

the standard MIM in SG13S have limited modelling of parasitiv effects. For RF/microwave, we added rf_cmim model to the PDKs, which has a more defined substrate path and a more detailed eqivalent circuit model.

The model parameters were extracted from my EM simulations using Sonnet. IHP can provide a document with more details and a comparison to measured rf_cmim data.

If you want to use Momentum, pay attention to these details:
- Set physical model for vias to 2D (not 3D) to enfore purely vertical current flow in merged via arrays.
- Instead of using my via merging with derived layers (merging to one large block -> changing cross section) use ADS2016 Preprocessor > Global Patterns with an approriate Via Simplifaction method that keeps the total via cross section
- For the mesh, you can set mesh size in absolute values (microns) instead of 1/nn wavelength. But accurate model for this MIM requires a detailed simulation, there is no magic solution to make it accurate and fast.

Best regards
Volker