Statistical Reliability of Phase Fraction Determination Based on Electron Backscatter Diffraction (EBSD) Investigations on the Example of an Al-TRIP Steel

The aim of this article is to discuss the representativeness of electron backscatter diffraction (EBSD) mapping data for phase fraction determination in multiphase materials. Particular attention is paid to the effect of step size and scanned area. The experimental investigations were carried out on a low-alloyed steel with transformation induced plasticity (TRIP) that shows a relatively heterogeneous distribution of residual austenite in a ferrite matrix. EBSD scans of various area sizes and step sizes were carried out and analyzed with respect to the determined austenite phase fraction. The step size has only an indirect influence on the results, as it determines the size of the investigated area if the number of measurement points is kept constant. Based on the experimental results, the optimum sampling conditions in terms of analyzed area size and the number of measurement points were determined. These values were compared with values obtained from Cochran’s formula, which allows calculation of sampling sizes for predefined levels of precision and confidence. A significant deviation of experimental from theoretical optimum sample sizes was found. This deviation is, for the most part, a result of the heterogeneous distribution of the austenite phase. Depending on grain size and volume fraction of the second phase, the false assignment of phases at grain boundaries also may introduce a significant error. A general formula is introduced that allows estimation of the error caused by these parameters. Finally, a new measurement scheme is proposed that allows improvement of reliability and representativeness of EBSD-based phase determination without large sacrifices in measurement time or data set sizes.