Within a full density functional theory framework we calculate the band structure and doping potential for phosphorus δ-doped silicon. We compare two different representations of the dopant plane; pseudo-atoms in which the nuclear charge is fractional between silicon and phosphorus, and explicit arrangements employing distinct silicon and phosphorus atoms. While the pseudo-atom approach offers several computational advantages, the explicit model calculations differ in a number of key points, including the valley splitting, the Fermi level and the width of the doping potential. These findings have implications for parameters used in device modelling.