|
The efficiency and accuracy of three-dimensional (3D) reconstruction and quantitative characterization of grain structures strongly rely on the section thickness of serial sections. Taking the single-phase polycrystalline structure generated by 400×400×400 Potts Monte Carlo simulation as the study object, influences of the section thickness on the measurement precision of grain size are studied. The number of grains contained in the structure are 948, and the voxel value (or volume) of individual grain in the 3D simulated microstructure varies from 3.98E5 to 1.52E2. The section thickness is represented by the relative values, i.e., the ratio between the section thickness (h) and the average grain intercept of the polycrystal (<l2>), or the ratio between h and the caliper diameter of the 3D grain (DT). The results indicate that, for the determination of grain volume and grain surface area of individual grain, the maximum allowable values of h/DT should be 0.15 and 0.10, respectively in order to maintain the requirement of the relative error less than 5%. On the other hand, the relative grain size R/<R> affects the relationship between the relative error and the value of h/<l2>. For instance, the values of h/<l2> should be less than 0.20 and 0.15, respectively, in order that the relative errors of grain volume and grain surface area to be less than 5%, when the relative grain size is between 0.5 and 1.8. Beside, when the relative grain size is larger than 1.8 or smaller than 0.5, the value of h/<l2> would be correspondingly larger or smaller. |
|
Keywords:3D reconstruction; section thickness; quantitative characterization; grain size; Monte Carlo simulation |
|