“…If the LPP and SPP 5f occupations obtained from Mulliken population analyses differ by less than 0.5 electrons, the application of the LPP should yield reasonable results. In the case of actinide fluorides, the 5f occupations differ by at most 0.1 electrons for AnF 2 /AnF 3 /AnF 4 (except for ThF 3 /PaF 3 , where the trivalent oxidation state is not preferred/unstable), 0.3 electrons for AnF 5 , and 0.9 electrons for AnF 6 [12], which explains the good results using the di-, tri-, and tetravalent, the acceptable results using the pentavalent, and the bad results using the hexavalent LPPs. Thus, the 5f-in-core approximation reaches its limit for the hexavalent oxidation state and therefore these LPPs should only be used to preoptimize structures.…”