Use of Rietveld refinement for elastic macrostrain determination and for evaluation of plastic strain history from diffraction spectra

Macrostrain variations in engineering components are frequently examined using neutron diffraction, at both reactors and pulsed sources. It is desirable to minimize the sampling volume in order to maximize the spatial resolution, although this increases the required measurement time. At reactors, macrostrain behavior is inferred from a single lattice reflection (deemed to be representative of the bulk response). At a pulsed source, a complete diffraction pattern is recorded and accordingly it is natural to fit the entire diffraction spectra using a Rietveld [J. Appl. Cryst. 2, 65 (1969)] refinement. This means that an idealized crystal structure is fit to the measured distorted crystal structure, which includes deviation of the measured lattice reflections from the ideal due to elastoplastic strain anisotropies, which are dependent on the particular lattice reflection (hk1) considered. We show that elastic macrostrains calculated from lattice parameter changes in Rietveld refinements (without accounting for hk1 dependent anisotropies) are almost identical to the bulk elastic response and are comparable to the response obtained from a single lattice reflection typically used by practitioners at a steady state source. Moreover good refinements on the complete pattern are obtained with short measurement times compared to what is required for good statistics for single reflections. By incorporating a description of the elastic strain anisotropy expected in cubic materials into the Rietveld code, an empirical prediction of plastic strain history is possible. The validity of these arguments is demonstrated by analysis of a uniaxial tensile load test and a reanalysis of previously reported data taken on a deformed stainless steel ring. The plastic strain predictions compare favorably with a finite element model. (C) 1997 American Institute of Physics.