Sorry for the tardy reply. Please find my response to your questions below,
Yes. Since you have a periodic array of nanoparticles, you should use a periodic FDTD. However, using the TFSF source is not going to be right choice in most cases (including your plasmonic solar cell). If you look at the “TFSF source in Periodic Boundaries” page I mentioned above, you will notice a comment at the top of the page saying “The TFSF source should only be used in situations where scattering in the specular (zeroeth order) direction is of interest”. In all other cases a plane wave source is the more appropriate choice.
Regarding your use of the scattering cross-section analysis group, I have discussed this with my colleague @kchow since I myself am not an expert on the topic. What I have learned is that the figure of merit “scattering cross-section” is more appropriate when simulating scattering from a single scatterer rather than a periodic structure. In such a case (single particle), since you will use PML boundary condition you will also use the TFSF source. In a periodic structure like yours, the more meaningful result is probably the grating order transmission which can be calculated using the corresponding analysis group. Please take a look at this example in KB (Grating order transmission) and let me know if you have any questions.