Dear Subhra, The course of action I would take is the following: Firstly make sure that the SIESTA calculation is as exact as possible. To this aim: - Take care that the basis set is the largest (DZDP or better) - That the MeshCutOff is large enough (300 Ry). - Also use relatively low EnergyShift 20 meV. - Use LDA and GGA to make sure the functional does not influence the order of the levels (GGA PBE should be better than LDA in this case). - Of course make sure that the pseudopotential used is the most accurate possible (relativistic correction included if relevant)
Then, if the experimental measurement does not coincide with the result several things could have happened: - DFT itself does not describe the system correctly. You could try to recompute the same system with another DFT numerical approach: A Plane Waves code suggested. This will probably not change much (SIESTA is good enough!), just to confirm that DFT itself is not "enough". - Last but not least, the theoretical model and the experimental situation may differ. Who knows, there might be some contamination or error in the experimental results interpretation? Some fields not taken into account? ... Best regards Jose A.
