Combining polarized light microscopy with machine learning to map crystallographic textures on cubic metals

In this work, we demonstrate the possibility of gathering crystal orientation information from cubic—optically isotropic—materials using polarized light microscopy. Our method relies on a simple and inexpensive optical technique called polarized reflectance microscopy (PRM). During PRM, we capture a sequence of micrographs of chemically etched aluminum samples under different polarization angles. We then feed the measurements into a machine learning algorithm that interprets light reflection intensity as a function of polarization angle and returns the location-specific crystallographic texture across the sample surface. We discuss the physical mechanism behind the connection between polarized light reflectance and crystal orientation in this class of optically isotropic materials, and the opportunities that such a technique would open in the field of high-throughput characterization of materials.