Reducing the effects of texture on phase fraction measurement of retained austenite using x-ray diffraction

Abstract

A method for more accurate phase fraction measurements of retained austenite (RA) using X-ray diffraction (XRD) has been developed and tested. RA presence improves formability, ductility, and dynamic strengthening in steel products. These properties are tied to the amount of austenite present in the material. As such, accurate phase measurement is key in understanding and predicting material behavior. One method for RA measurement is XRD. However, results can exhibit texture bias error caused by preferred crystallographic orientations. Sampling schemes have been considered as a means of reducing texture bias error. Sampling schemes designate a series of tilt and rotations to define diffraction vectors for use during the XRD measurement. In this study, sampling schemes with vectors arranged in a hexagonal (hex) grid pattern over the pole figure were investigated. Simulations were conducted using texture data of TRIP 700, TRIP 780, and duplex stainless sheet steels to evaluate the hex schemes and compare to more typical normal direction (ND) measurements. Hex grids with multiple grid spacings and different tilt limits were used to determine the number of vectors and pole figure coverage necessary for an accurate measurement. The “30-p” simulated hex scheme only contained 4 diffraction vectors and was limited to 60° of tilt, yet exhibited accurate results. The ND results were relatively inaccurate for all four materials. Further simulations using texture component combinations were performed to gauge the robustness of the sampling schemes. 7 ferrite texture components and 13 austenite texture components were produced using the Matlab package MTEX. A simulation was conducted using each sampling scheme and texture combination. The results indicated the full hex schemes could adequately measure materials with very sharp or unknown texture profiles. The hex schemes reduced bias error over the ND simulation results. Validation of the duplex sheet steel simulations was conducted using experimental measurements. Multiple peak combinations were used to determine the effect of peak selection on accuracy. Both ND and 30° partial hex (comparable to the 30-p simulation) measurements were performed. The hex results were largely within one standard deviation of the accurate value. The ND measurement results remained less accurate. The simulation results for these same sampling schemes strongly agreed with the measurements.