Spectral Analysis of Correlated One-Dimensional Systems with Impurities

Abstract

An averaging procedure is proposed to account for spectral features of correlated one-dimensional systems in the presence of non-magnetic impurities. The dynamical spin structure factor for a corresponding random ensemble of Heisenberg chain segments is calculated by exact numerical diagonalization. It is shown that a few-pole approximation is sufficient to describe the numerical results. A similar analysis is proposed for the discussion of experimental spectra, such as obtained by inelastic neutron scattering measurements on Zn-doped CuO chains. By examination of the disorder-induced pseudo-gap, the loss of spectral weight, and the discrete peak structures due to smallest-cluster contributions, the underlying impurity distribution function can be determined.