How to measure loss tangent value experimentally?

For example, I have a silicon substrate with a known resistivity value. Can someone help to describe the experiment requirements or procedure?

both tanδ and conductor loss are responsible of the dissipative attenuation which is the Available Gain.

1st measure spar

2nd calculate available gain

3rd into a microwave simulator like ADS or Microwave Officie or Genesys put a microstip line of the same phisical geometry and put the know metal resistivity, and find iteratively what value of tanδ satisfy the calculated value of Gav.

4th You may take advantage if you analize a function vs. frequency. Mathematically find coefficents on the empirical function on the following form:
Gav[dB]= C0*Freq^0.5+C1*Freq^1
C0 is the metal loss while C1 is the dielectric loss.

Note: it's important that the measured spar be inside the frequency range into that the Zo is a constant. For lossy lines, especially at lower frequency, Zo may be different. If you cannot know this phenomena, post me a request for the Agilent AN on this topic.

Thank you, sergio mariotti.

I not fully understand the principle in your description. Could you give me a reference book or paper name?

This will be an interesting experiment. How about I make a CPW on the wafer, then measure the spar, and use ADS to simulate with the same dimension?

I like to build a microstrip resonator, measure the passband characteristics, model it in a simulator (I assume the conductor loss model in the simulator is accurate) and vary the simulated substrate loss tangent around until the results match.

maybe you should form a resonator with the material filled inside.and then you can test a Q factor.
Q=1/(1/Qc+tanσ)
if Qc is calculated then tanδcan be achieved.

http://www.emclab.umr.edu/pdf/TR00-1-041.pdf

tnx for 3 points.
The law Gav[dB]=C0*F^.5+C1*F may by found on many sources.
I.e. Application Notes on Huber+Suhner-RF Cables catalog.
or Compact Software (now Ansoft) SuperCompact (now Harmonica) 5.5 User Manual-element: TRL
The phisycal meaning is: conductor loss is dominated by skin effect wich is related to the square of frequency. I don't know a phisycal meaning that explain dielectric loss.

you can also fabricate an slice with the cross section the same as a wave guide corss-section.and put it in the wave guide terminated with short circuit and open circuit,and the 50 terminator.Measure the S11(vector) respectively and then deduce the Eps.