PML boundary conditions are incompatible with plane wave sources. Use a Gaussian source focused on your absorbed power analysis group or use Bloch/Periodic or Symmetric/Anti-symmetric BC’s with the plane wave. Also the simulation is not reaching auto-cutoff, meaning that their is still energy bouncing around in the slab. Ending the simulation before the energy is allowed to decay will create artifacts in your analysis. Consider reducing the z-span and/or increasing sim time.
Since this is a 3D (+ 1D k-space) analysis group make sure that you are monitoring the power in Si slab by changing the z value of the plotted slice using the sliders at the bottom. Compare this to the Pabs that you would expect from the material properties alone, but the geometry will have resonances that could change the results. The script command stackrt may be appropriate for 1D problems like this?