Inconsistency of band lineup with commercial software

[duarte-jfs]

Hi, I am trying to leverage solcore's sesame PDD for my project, and to make sure everything is working as I expect, I decided to benchmark it to NextNano. My benchmark is a PIN diode consisting of P-InP | InGaAsP (As=0.53, lattice matched to InP) | InP - N. Almost all of the outputs are matching very well for my degree of accuracy needed. However, I notice that in sesame, the valence band barely moves while the conduction band moves much more than in the NextNano solver. Such offset is causing a charge accumulation at the boundary and consequently a big rise in the electric field. This effect will have a big impact on my future calculations, and therefore I would like to understand what can be done and which of the results should we trust more.

Notice that the bandstructure output of the sesame has been shifted to match the one from nextnano (output_sesame['some_band'] = output_sesame['some_band'] - min(output_sesame['Ec'])

Could you help me understand this discrepancy?

PS: Great work on the package! I am very happy that there is now an open-source alternative for poisson-drift-diffusion simulations