Using non-parametric statistical testing to quantify solute clustering in atom probe reconstructions

Area of Science:

• Materials Science
• Nanotechnology
• Statistical Analysis

Background:

• Atom probe tomography (APT) is crucial for analyzing nano-scale solute distributions in complex materials.
• Quantifying solute clustering in APT data lacks standardized methodologies.
• Comparing different APT datasets is challenging due to limitations in analyzing nearest-neighbor distributions.

Purpose of the Study:

• To develop a quantitative method for assessing solute clustering in APT data.
• To introduce a statistically robust approach for comparing APT datasets.
• To overcome limitations of conventional comparison methods like the Pearson coefficient.

Main Methods:

• Adaptation of a non-parametric hypothesis testing statistic, 'the level of heterogeneity', for nearest-neighbor distributions.
• Utilizing Monte-Carlo simulations to calculate the heterogeneity statistic.
• Developing a novel method for comparing APT datasets based on Monte-Carlo simulations.

Main Results:

• The 'level of heterogeneity' statistic effectively quantifies solute clustering beyond random distributions.
• The proposed Monte-Carlo simulation method provides a more powerful approach for comparing APT datasets.
• The new method surpasses the efficacy of the conventional Pearson coefficient for dataset comparison.

Conclusions:

• The developed statistical framework offers a standardized and powerful approach to solute clustering analysis in APT.
• This method enhances the reliability and comparability of nano-scale solute architecture investigations.
• The findings advance the quantitative analysis capabilities within materials science using APT.