The singularity at the Fermi edge in the soft-x-ray spectra of metals is known to be described by a factor ${|\frac{{\ensuremath{\xi}}_{0}}{\ensuremath{\epsilon}}|}^{\ensuremath{\alpha}}$ multiplying the one-electron intensity. Using a separable potential, Nozi\`eres and deDominicis showed that $\ensuremath{\alpha}$ is a function of the Fermi-electron phase shifts. However, ${\ensuremath{\xi}}_{0}$, treated until now as a constant, has not yet been derived. We extend the above factor to other frequencies, writing it in the form $G(\ensuremath{\epsilon}){|\frac{\ensuremath{\xi}(\ensuremath{\epsilon})}{\ensuremath{\epsilon}}|}^{\ensuremath{\alpha}}$. Using simplified diagrams we can introduce a realistic potential and orthogonalized plane waves and we determine $G(\ensuremath{\epsilon})$ and $\ensuremath{\xi}(\ensuremath{\epsilon})$ in a range of about 3 eV from the edge. The calculations are applied to $\mathrm{Na}{L}_{2,3}$, $\mathrm{Li}K$, and $\mathrm{Be}K$ bands. The factor $G(\ensuremath{\epsilon})$, related to the open-line part of the problem, presents a singularity in the slope at $\ensuremath{\epsilon}=0$. This fact, important in the $K$ bands, was not realized before. However, the edge singularity does not appear to be strong enough to explain the premature peak in the $K$-emission bands. The $p$-scattering resonance discussed by Allotey is probably dominant here.
Read full abstract