Changes in density of states caused by chemisorption, with implications for photoemission

Abstract

Considerable effort has recently been devoted to investigating the changes in photoemission from transition metals when light gas atoms (H, 0, N, C, etc.) are absorbed on their surfaces 1, 2). We can gain insight into the features of the changes by investigating the change in electronic density of states (A DOS) using the framework presented by Einstein and Schrieffers) (hereafter denoted by ES). This one-electron model provides a simple way to visualize the formation of peaks below the d-band, the depletion of states within the center of the band, and the general dependence of spectra on the natural parameters of the problem. The model of ES considers the (100) surface of a single s-band, simple cubic, semi-infinite lattice in the tight-binding approximation [(loo) cubium], as treated by Kalkstein and Sovend), and Allans), with the one-center matrix element set at zero; the center of the bulk band is the energy zero. We take the two-center matrix element to be -3, which sets the energy scale and gives a bandwidth of 6. We represent the adatom by a single, sharp, nondegenerate (save for spin) level of energy E,. The best rough estimate of E, is an average of the ionization (I) and affinity (A) levels, rather than simply the former. The model is most realistic for neutral adsorption, i.e. when E, and the Fermi energy are relatively close, and when the Coulomb interaction (-Z-A) is not too large. The perturbation parameter, V, characterizes the hopping of an electron between an adatom and its nearest neighbor on the substrate, as in the Anderson model6). For clarity and brevity, our presentation assumes that the adatom sits directly above a surface atom in the atop position. The present discussion neglects self-consistency effects; we are presently implementing them’). We note here, however, that for atop binding, particle-hole symmetry gives self-consistency automatically for a halffilled band and E, = 0. We will therefore keep E, near the band center even though this may tend to overemphasize adatom-substrate mixing.