The major problem of your simulation file is that the simulation time is not long enough, in fact way shorter than what is required. I believe your device has a super high Q factor - very strong in trapping light. This makes it necessary to use a simulation time long enough to make sure light has already decayed. (the auto shutoff value can give you some idea). Otherwise, you may see frequency domain data larger than 1 since the time domain data is not complete.
I have edited your file Large PC_Small Polymer Test.fsp (513.7 KB)
and this is the new reflection. Although it still has some values slightly larger than 1, I think it is much better now. You can further increase the simulation time (now is 100x than what you had), and play around with mesh to refine the results.
I also noticed that this is a periodic device so I have reduced the simulation span in such a way that it only contains one unit cell. This will take the periodicity in the consideration at the same time makes a much faster simulation. I have also applied the BC symmetry and mesh to best speed up the simulation but being able to resolve the fine features you had, based on my best guess.