Multivariate ALCHEMI in the presence of anti-site defects

It is shown that a recently developed formulation of ALCHEMI, which takes into account the x-ray signal delocalisation effects, is not suitable for adaptation to the case of anti-site defect density determinations. However, the multivariate formulation of the method ALCHEMI is then applied to an Ni-Fe-Cr ordered alloy sample containing significant numbers of anti-site defects, and whose site distributions for all three elements were known in advance. In this way the importance of the presence of such defects is quantitatively determined for a real case. It is found that for this sample, the errors introduced are less than those arising from other sources, and it appears likely that in general the method can tolerate fairly high defect levels for compounds of host stoichiometry AxB where x is 3 or higher. (orig.)