145HIGHLIGHTS 2020

boundary properties. The fitting methodology is illustrated in Figure 126: it combines a 3D experimental movie of the grain growth with 3D phase-field simulations. The reduced mobilities m = {m1, m2, m3, , mn}, where n is the number of grain boundaries in the system, are iteratively refined until the simulation optimally matches the experiment.

The 3D experimental movie was generated by diffraction contrast tomography (DCT) at

Fig. 126: Schematic diagram of the fitting method. The 3D position of grain boundaries, uexp(t), is determined experimentally as a function of time, t. Using one time-step t0 of the experimental movie as the initial configuration, a 3D simulated microstructure usim(t, m) is generated by a phase-field model with an initial guess of the reduced mobility mi of each boundary i. At a later time-step tfit, the simulated microstructure is compared with the experimental one, and a cost function fcost is defined to quantify the difference between the experiment and the simulation. The cost function is minimised to find the set of reduced mobilities mfit of each boundary that provide a best match between the simulation and the experiment.

beamline ID11. The orientations of all grains and the grain boundary positions were mapped with a resolution of 0.1 degrees and 1.5-3 µm, respectively, during annealing at 800oC. The movie comprises 15 time-steps spaced by five minutes. A standard phase-field method was used to simulate the grain growth. By fitting with 35 combinations of time-steps (t0, tfit), 10,307 reduced mobilities of 1,619 internal grain boundaries were determined. Figures 127a-c illustrate the quality of the

Fig. 127: Comparison of the experimental and the simulated grain boundaries within one slice of the 3D volume, (a) with the initial guess of the mobilities and (b) with optimised/fitted mobilities. c) The morphology evolution of one grain as determined experimentally and simulated with the optimised reduced mobilities. The logarithm of the reduced mobility as a function of (d) misorientation and (e) grain boundary inclination shows no strong correlation

between mobility and the five crystallographic degrees of freedom of the grain boundary.