Multivariate Statistical Analysis of Atom Probe Tomography Data

Multivariate statistical analysis (MVSA) methods have been shown to yield many advantages in analytical instrumentation techniques because they offer an objective means to find complex information correlations in data that are not obvious to an observer and they maximize the discrimination of low-level signals from noise. Because they are objective in nature, the methods hold promise for automated data mining, especially in large data sets. Atom probe tomography (APT) hardware improvements in the past decade have shifted the bottleneck from data collection to data analysis. It is imperative, therefore, that methods are developed to automate analysis of the entire APT data set. MVSA has been applied successfully to other types of large 3D tomographic data sets including EDS [1] and ToF-SIMS [2]. We report, here, the adaptation of those techniques, in particular multivariate curve resolution (MCR), to APT data.