“…We therefore adopt a single-scattering model, assuming that the sample is thin enough to make plural inelastic scattering unimportant and that the effect of elastic scattering can be dealt with separately as beam broadening (as just discussed). In the case of crystalline specimens and core-loss spectroscopy, these assumptions become questionable and a proper treatment requires a simultaneous treatment of elastic and inelastic scattering, using (for example) a combination of multislice and single-inelastic scattering theory [21,22,24] or Bloch-wave methods [18][19][20].…”