A machine learning approach to map crystal orientation by optical microscopy

Abstract

Mapping grain orientation in crystalline solids is essential to investigate the relationships between local microstructure and crystallography and interpret materials properties. One of the main techniques used to perform these studies is electron backscatter diffraction (EBSD). Due to the limited measurement throughput, however, EBSD is not suitable for characterizing samples with long-range microstructure heterogeneity, nor for building large material libraries that include numerous specimens. We present a machine learning approach for high-throughput crystal orientation mapping, which relies on the optical technique called directional reflectance microscopy. We successfully apply our method on Inconel 718 specimens produced by additive manufacturing, which exhibit complex, spatially-varying microstructures. These results demonstrate that optical orientation mapping on a metal alloy is achievable. Since our method is data-driven, it can be easily extended to different alloy systems produced using different manufacturing processes.

Quantitative multi-image analysis in metals research

“…

Figure 2 DRM workflow (adapted from Refs. 19 – 21 ). (a) A typical DRM apparatus includes a stereo-microscope and motorized stage, which is used to control the illumination direction (parameterized by the elevation, , and azimuth, , angles).

…”

“…[ 22 ] When the interpretation of the optical signal becomes more challenging—for example in multi-phase metal alloys—machine learning (ML) can establish the links between directional reflectance and crystal orientation. [ 21 ] Both types of algorithms are able to yield grain orientation maps that are essentially equivalent to those provided by electron backscatter diffraction (EBSD), [ 23 ] but over much greater sample areas and without having to place the sample in a vacuum chamber.…”

“…DRM has been demonstrated on pure metals, [ 20 , 24 ] semiconductors, [ 19 ] metal alloys, [ 21 ] and ceramic reinforced polymer composites. [ 25 ] However, it holds greatest promise wherever conventional characterization techniques, such as EBSD, struggle to assess large-scale microstructure variability, e.g., in additively manufactured materials.…”

Machine learning enables precise holographic characterization of colloidal materials in real time

“…Machine learning algorithms are revolutionizing measurement science by decoupling quantitative analysis of experimental data from the mathematical representation of the underlying theory. 1,2 The abstract representation of a measurement principle that is encoded in a well-designed and well-trained machine-learning system can rival the precision and accuracy attained by fitting to an analytic theory and typically yields results substantially faster. Gains in speed and robustness have been particularly impressive for measurement techniques based on video streams, [3][4][5][6] which typically involve distilling small quantities of valuable information from large volumes of noisy data.…”

“…20,21 Hologram analysis is a challenging inverse problem 7,22 both because recorded intensity patterns necessarily omit half of the information about the light's amplitude and phase profiles and also because the underlying Lorenz-Mie theory of light scattering is notoriously complicated. [23][24][25] Extracting quantitative information from holograms is an unusual application for machine learning in two respects: (1) it involves regression of continuously varying properties from experimental data and (2) the machine-learning system can be trained with synthetic data generated from an exact theory. 4,10,26 The trained system therefore embodies a simplified representation of the underlying theory over a specified parameter domain that can be computed rapidly enough to be useful for real-world applications.…”