The mystery of missing species in atom probe tomography of composite materials

There is a serious problem in atom probe tomography of composite materials such as oxides that even from stoichiometric samples one observes non-stoichiometric ion yields. We present a quantitative model that explains the non-stoichiometry allowing a fit to experimental data of ion yields as a function of applied field to extract activation barriers and prefactors. The numbers are confirmed by density functional theory. We also show that for oxides the missing oxygen is thermally desorbed as neutral O2, either directly or associatively. Finally, we suggest methods to improve the experimental setup.