sp332: If you are still reading, thanks again for the link. I haven't delved deeply, but I believe the paper is aimed at exploring whether graphene's photocur

sp332: If you are still reading, thanks again for the link. I haven't delved deeply, but I believe the paper is aimed at exploring whether graphene's photocurrent mechanism is in fact thermoelectric in nature. If you look at Eq 1 in the arxiv paper, you will see that they start by calculating the Seebeck coefficient from first principles. (My original comment above turns out to be exactly the issue.) The reason that this is interesting: The authors are saying that there is an interesting Seebeck effect as a consequence of graphene's material properties, especially its odd density of states. To the point you raised: The electron population does absorb photons. But rather than transporting across the pn junction due to the increased energy of the absorbed photon, it collides with other electrons and then these electrons form a population with an elevated temperature among themselves. The electrons do not couple with the lattice, so they don't transfer energy to the lattice all that quickly. Hot electrons from the high-temperature ensemble then have a probability of making it across the pn junction. Like I said above, I haven't fully digested the paper (and I doubt I will, it's not in my critical path), but that was my quick take. Any other readers, please correct me at will.