haydar shalom,
How are you doing
1)
Copper skin depth in 100 Mhz is 6.7um.
Copper skin depth in 67 Mhz is 8.18um.
If you are using copper thicker than 25um than the losses should be calculated correctly in both frequencies. Otherwise you will need to solve inside the metal, or use the special skin depth meshing.
2) There could be impedance mismatches. When comparing the S-parameters always compare also Z0 (assuming waveport solution). If you are using a lumped port then Z0 will stay the same but S11 will be different because of an impedance mismatch.
3) You can always use 100Mhz for mesh adaptation and then add a single point frequency sweep on 67 Mhz. This is the best way to compare 100 Mhz and 67 Mhz results.
Regards,
Itai