How to measure capacitance after fabrication

[mordak]

Hi,

I have some MIM capacitors in my circuit and I want to find the real value of them after fabrication. So I was planning to put some dummy caps close to the main caps and then measure capacitance of the dummy ones to have some idea how much error I have (I already know the maximum error according to what the fab company reported, but I want to have the real error). Considering parasitic cap of the pads and other parasitic sources, what is the best way to measure the fabricated cap?

Thanks

 

[cl10greg]

If available, an LC meter is the best way. If you had to rig something up you make your own LC meter by using either a voltage or current measurement method. You could also do the frequency method like done with LC meters.

 

[mordak]

If available, an LC meter is the best way. If you had to rig something up you make your own LC meter by using either a voltage or current measurement method. You could also do the frequency method like done with LC meters.

I guess you're talking about using LC meter to measure the capacitance, if it's the case, like I said due to the capacitance of the pad which is away greater than the MIM cap I'm going to measure, I can't directly connect the cap to any pin.

 

[dick_freebird]

What kind of measurement equipment do you have?

A GSG probe and S-parameter measurement equipment are
certainly capable of pretty low value capacitance readings,
especially if you have an opens and shorts de-embed pair
on the same die.

You might also want to make arrays of unit MIM caps and
do the math, rather than try to measure just one. That
will help bury the interconnect capacitance somewhat.

Don't blame the pads for inability to measure; put down
a duplicate structure, delete the MIM cap, take two
readings and do the math. You also don't have to use
approved pad structures with all their baggage; a passivation
cut and top metal only makes a perfectly good probe pad.
If this is a test chip and not something that needs to
pass some pile of corporate a$$-covering can't-do rules,
anyway.

 

[phongphanp]

I had found topic on DC-DC converter analytic to inspect the some damage point , It concern more Mathematics Questions. in this your measuring , the indirect method seem morely preserve the important charecteristics.

 

[mordak]

What kind of measurement equipment do you have?

A GSG probe and S-parameter measurement equipment are
certainly capable of pretty low value capacitance readings,
especially if you have an opens and shorts de-embed pair
on the same die.
.

I don't have any GSG probe, and no probe station, the only equipments I have access to would be scope, spectrum analyzer and simple stuff like this. But I'll try having some unit caps in parallel to get the bigger one.

put down a duplicate structure, delete the MIM cap, take two
readings and do the math. You also don't have to use
approved pad structures with all their baggage; a characteristics
cut and top metal only makes a perfectly good probe pad.

This is a test chip and just for research, so I'm not looking for anything awesome or precise, just to get a feeling that how off I am from what I expected. I'm not sure if two pads would have the exact same characteristics or not, I mean if I duplicate the structure, there might be some differences between pads used for these two test structures. Say there is 200fF difference between two pads, it might be way larger than the unit cap I have, I have no idea if I can assume two pads are exactly the same or not, can I?

- - - Updated - - -

I had found topic on DC-DC converter analytic to inspect the some damage point , It concern more Mathematics Questions. in this your measuring , the indirect method seem morely preserve the important charecteristics.

Hey, frankly I didn't get it, is this about the measurement I was asking?

 

[phongphanp]

I mean about you should try some operation mode of some architecture.

 

[jiripolivka]

Be sure the LC meter you intend to use would not damage the tested MIM structure. Use capacitor coupling and make sure the test signal voltage is small enough compared to voltages in the tested circuit.
The coupling capacitor should be larger than the expected MIM capacitor or inductor impedance at the test frequency.

 

[phongphanp]

All above is for basic , try Network Analyzer to make the test, s-parameter and read the manual about on this topic.

The de-inc impedance is just the approximation for HF.

 

[D.A.(Tony)Stewart]

Normally the biggest variance is etchback factor and dielectric constant. TE's will usually design test coupons in board test shops for TDR testing to measure impedance of test pads or tracks for transmission lines tests outside the useful board outline area to ensure the buried tracks are compliant. THis adds cost to the board to include impedance testing with TDR but useful for tighter tolerances.

 

[phongphanp]

Try search EDN at the Circuit Idea section .