Exploring Biases in Atom Probe Tomography Compositional Analysis of Minerals

Abstract

Atom probe tomography (APT) is emerging as a key nanogeochemistry technique for diverse geoscience applications. Estimating stoichiometric mineral compositions is particularly challenging for features at nanoscale. APT provides reliable measurements for metallic systems but the reliability for oxides is problematic, notably due to oxygen deficit. Here, we use laser‐assisted APT to compare results for spinel and garnet crystals. APT compositional results were compared with those by electron microprobe to determine the possible APT analytical inaccuracy. Extensive data processing was accomplished, including correlation histograms, 2D ion distribution maps and 1D elements concentration profiles, to disclose the possible mechanisms leading to mineral stoichiometry biases. Multiple events and neutral molecules formation are probable the main processes responsible for atom deficit. In particular, the amount of the same isotope–same charge state ion pairs correlated with aluminium and oxygen deficits suggests that the co‐evaporation in a dead space–dead time window could lead to a significant decrease of detected ions. Also, molecular species dissociation and direct current evaporation could partially account for further atom loss. Overall, better APT compositional estimation was obtained for spinel, which has lesser variation in lattice sites and greater overall lattice symmetry, higher thermal conductivity and lower band gap compared with garnet.