site:www.edaboard.com hfss capacitor
The Si-Sub is low resistivity( 3~8Ohm*cm), a 2um Oxide is formed before the MIM
Cap fabrication(CMOS Backend Process). The insulating layer of MIM is Silicon Nitride, thickness on the ordor of 0.2~0.5um. Two metal plate is Copper.
But in HFSS simulation, I found strange results especially in Q calculations. If I reduced the thickness of two metal plates to 0.1um, the Q results are approximately equal that of 2um thick copper metal case. Any models published on articles would not agree.
I guess this error arises from the simulation setup, can anyone give some tips to simulate this structure accuratley? How to choose mesh and sweep
In view of the computing power, I currently tried interpolating sweep, which is similar to an Spiral Inductor expample in HFSS document.
An early post here mentioned that SONNET might be better for MIM simulation, but my final goal is to simulate and MEMS LC filter structure which mainly a 3D.
Added after 2 minutes:
I forgot to say that , the strange Q factor results remain even if I simulate this MIM structure surrounded by air( exclude silicon substrate effects).
Thank you !
Using hfss for MIM quality factor calculation seems like using a bomb for opening a shoe lace. Why don't you use ADS it uses moment method and answeres much faster than HFSS. I think ADS is the most perfect software for what you are going to do.
Mdanaie's "bomb" statement is beautiful, nice one! I agree.
In this case, Wizardz, I think you should consider using a combination of tools. Momentum should be OK for the MIM cap. Sonnet will do it better (I do work for Sonnet, but I am confident that my statement is true), but either tool should be up to the task.
You could use HFSS for the arbitrary 3-D part, but the problem there is that there is no supported design flow between Ansoft HFSS and either Momentum or Sonnet. If you start doing this on a regular basis, and a file format changes somewhere (some one added a semi-colon in some file format, for example), your entire design flow comes crashing down until you figure out what happened.
If that might be a factor, you could consider Agilent HFSS. I have not seen any comments about Agilent HFSS, so I do not know how it compares technically with Ansoft HFSS. However, I am sure the design flow to ADS will be supported, and then there is a well established and supported bi-directional design flow between ADS and Momentum and Sonnet. In fact, we often recommend designers double check results by analyzing in two simulators. It is unlikely that two simulators (in this case Momentum and Sonnet) will give the same error.
Another possibility is to simulate the 3-D portion in CST Microwave Studio (my company represents CST in North America). Supported design flow exists between CST, ADS, and Sonnet.
You can also try posting your problem on the Sonnet forum (www.sonnetsoftware.com) and see what people say. There are some world class experts that keep a close watch there, and the forum has become quite popular for that reason. Good luck!
Thank you, rautio!
I've read some of your posts, very helpful to me.
I can imagine what mdanaie meant by that "bomb" analogy. The problem is that "the bomb can't open a shoe lace".
I'll have to find the cause of that , may be try SONNET and other software is a good way.