Hi. In this particular structure, the key result that we are interested in is the optical absorption and the contribution of the metal to this result is negligible. However, if we use a real metal (Al) in the simulation instead of PEC then the simulation uses a much finer mesh over the metal (since the optical wavelength is much smaller inside the metal due to its higher index). The whole simulation then becomes extremely slow and requires a lot of memory. This is why we have used PEC in this example.
Now, the shortcoming of using PEC in place of a real metal is that PEC will reflect 100% of incident light whereas a real metal like aluminum might let a small fraction of the light to pass through. Also, if the purpose of the metal is to create surface plasmons then PEC will be unsuitable since PEC will not create surface plasmons. In the VPD example, none of these issues are of concern so we should be able to use PEC instead of Al here.
As mentioned earlier, this has to do with the large index of the real metal. However since PEC is a special material, the FDTD solver can handle it in a special way and it does not need to use a fine mesh for PEC.
I did a quick test and I noticed that using Al instead of PEC does result in a small increase in the photocurrent (15% increase) if I run the simulation with the current settings (I did use a coarse mesh in Al though to make the simulation fast). However I believe this happens due to the following reason:
In the simulation file, you will notice that the generation rate analysis group extends slightly inside the metal. In case of Aluminum some light goes into the metal and gets absorbed and therefore this additional absorbed light when gets included in the calculation of photogenerated current results in a larger current. However, in reality any absorption in the metal does not contribute to photogeneration of electron-hole pairs so even though the result in FDTD gives us a larger photocurrent the overall responsivity should not increase much.
When I made the analysis group smaller in Z-direction so it does not go inside the metal, the different in current between PEC and aluminum came down to a mere 2%. This variation is quite insignificant and I believe it shows that we can get accurate result with PEC while making the simulation faster.