An electron-gas model of a bimetallic interface is constructed by joining two semi-infinite half-planes of unequal positive charge and adding electrons until a charge-neutral system is achieved. The electrostatic (dipole) potential equalizes the Fermi energies of the high- and low-density components of the junction by compensating for the difference in their (bulk) exchange and correlation separation energies. The model is used to calculate self-consistently the charge density and one-electron (junction) potential in the region near the interface. The barrier height $V_b$ associated with the self-consistent potential is not related to the vacuum work functions ${\phi }_L$ and ${\phi }_R$ of the two-component metals by $V_b\ne {\phi }_L-{\phi }_R$. This result is due not to either real or "virtual" surface states, but rather to the redistribution of the electronic charge at the biometallic interface relative to the vacuum interfaces of the separate metals. Localized "surface states" can occur for certain junction potentials on the high-density side of the interface. These states do not occur in the self-consistent potential for the numerical example of $n_L={10}^{22}$ and $n_R={10}^{21}$ ${\mathrm{cm}}^{-3\mathrm{}}$. In addition, $V_b\ne {\phi }_L-{\phi }_R$ for these positive charge densities. Although the electron density exhibits Friedel oscillations on both sides of the junction, only two of the oscillations on each side are explicitly incorporated into the model charge density used in the self-consistent loops. A semiclassical (Schottky) model of the depletion region on the low-density side of the interface is inadequate because of the presence of large evanescent contributions to the electron density.

• Received 15 March 1967

DOI:https://doi.org/10.1103/PhysRev.160.541