integrated inductor Q measurement

When I do a probe test for my inductor(formed by two metal layer and a via trench), it seems that the Q value is much lower than HFSS result and the L value is normal, the most critical thing is the peak Q freq shift from 4G of simulation to 1.7G of test, what will cause that?
PS: the inductor is about 600 pH and the metal width is about 60 um.[/img]

Metal width 60um? Is that your bondpad?

Is the bondpad included in the HFSS simulation?

The pad is not included in the EM simulaiton. But in the test, we have the test structure to eliminate the effect.

EM simulator cannot be so wrong, unless your simulation or substrate information is wrong.

If everything is ok, then it must be inaccurate deembedding.