The Rigaku Journal Vol. 22 / No. 1 / 2005, 16–25

Since its inception over 35 years ago, Rietveld refinement (Rietveld, 1969) has come a long way. In the beginning, the aim was to carry out crystal-structure refinement using neutron diffraction. As its primary purpose was refinement of the perfect crystal structures of polycrystalline samples, there was a need to introduce corrections to peak intensities and positions, as calculated for a perfect powder standard, in order to match experimental patterns affected by sample texture and defects (i.e. residual or microscopic strains/stresses, crystal defects, small coherently diffracting domain size, etc.). It was not until the 1990s that an effort was made to determine microstructural properties through Rietveld refinement. Popa (1992) worked out a procedure for texture determination by the generalized spherical harmonics, which was implemented into GSAS (Larson & Von Dreele, 2001) by Von Dreele (1997). Furthermore, an anisotropic line-broadening modeling for arbitrary crystal symmetry was proposed independently by Popa (1998) and Stephens (1999) which was implemented shortly thereafter into most major Rietveld programs. Peak shifts were used for the residual strain/stress determination by Ferrari & Lutterotti (1994), Daymond et al. (1997) and Balzar et al. (1998). Next, an interest in determining an accurate crystallite size distribution led to work on introducing several common size distributions (i.e. lognormal, gamma, etc.) into Rietveld refinement and similar fullpowder-pattern approaches (Krill & Biringer, 1998; Langford et al., 2000; Popa & Balzar, 2002). Recently, attempts are being made to include modeling of crystal defects and dislocation configurations (Ungár, 2001; Scardi & Leoni, 2002). These developments gave us new tools for assessing the microstructure and thus a complete characterization of samples. Although the main intention of these models is to allow the determination of microstructural properties, in general, they are more robust over earlier approaches that were devised only to correct for undesirable effects, such as relative peak intensity changes due to texture or anisotropic peak broadening due to strain effects. The aim here is to review the modeling of microstructural properties in Rietveld refinement and to contrast it with a traditional approach for evaluating microstructural properties. A few examples in the determination of the crystallite size distribution and residual strain and texture will be given, in order to illustrate some of the possibilities. This is not meant to be an exhaustive review of the topic, as the main intent is to focus on our previous work. Moreover, the microstructural applications of Rietveld refinement are quickly developing into a powerful tool with a broad range of physical effects that are being modeled, which would be hard to cover in detail here.