Simulation of metamaterial with Periodic Boundary Conditions

I want to redo this article:
?Dynamical Electric and Magnetic Metamaterial Response at Terahertz Frequencies?
but I have find a big problem in the value of S11 and S21. I plot the S parameter which oscillating up and down. According to article, the sample is a planar array of SRRs fabricated upon semi-insulating gallium arsenide substrate (A planar array of SRRs are fabricated from 3 _m thick copper on a 670 μm thick high resistivity GaAs substrate. The outer dimension of an individual SRR is 36 μm and the unit cell is 50 μm). I believe simulation for free space should be carried out using master/slave BCs. I don?t know what?s wrong with my setup?

Here is the article:

Here is my project and result:

I removed the SRR and simulated the S parameter of GaAs substrate. the curve which I ploted also show oscillating up and down,but have a little shifting. The oscillating should be derived from the transmiting of wave in substrate.But I don't know why it is. who can explain this result?

Maybe A substrate with 670 μm is too thick? how should I choice the thick value of substrate in my model?

How do you obtain Transmission Spectra?

It is simple in hfss. Perform my hfss project in you PC,you will get.

posted your results so I think that you sould redo because it is not true.

Could you tell me what's wrong with my project or my model?
Thanks.
I don't consider the lossy dielectrics in higher frequency in my calculation,because I don't know how to defining the frequency-dependent material properties according to Drude model in hfss. I suppose it may be one of the reason for my inaccurate result.