Influence of indentation size and spacing on statistical phase analysis via high-speed nanoindentation mapping of metal alloys

The development of high-speed nanoindentation has enabled the acquisition of mechanical property maps over square millimeters of area with micron-scale resolution in reasonable amounts of time. This provides rich datasets which contain morphological and statistical data on the variation of mechanical properties in a microstructure. However, the influences of the indentation size and the deconvolution method employed on the extracted phase properties remain unclear. In this work, a range of depth/spacing increments was explored on two different materials systems, an Al-Cu eutectic alloy and a duplex stainless steel, representing an ‘easy’ and a ‘hard’ case for statistical deconvolution, respectively. A total of ~ 500,000 indentations were performed. A variety of statistical analyses were then employed and compared: the 1D analysis of Ulm et al. using 2 and 3 phases, a 2D rotated Gaussian fit, K-means clustering, and a visual comparison to 2D histograms. This revealed several different sensitivities of the deconvolution methods to various types of error in phase identification. Graphic abstract