Local Porosity Measurement From Scanning Electron Microscopy Images in the Backscattered Electrons Mode

Abstract

Abstract

SEM Image Processing Assisted by Deep Learning to Quantify Mesoporous γ-Alumina Spatial Heterogeneity and Its Predicted Impact on Mass Transfer

“…Extrudates were then cut into two sections, as shown in Figure a, and each section was impregnated under pressure in a mixture of liquid methyl methacrylate (MMA, Sigma-Aldrich) and azo-bis-isobutyronitrile (AZDN, Sigma-Aldrich), used as a radical polymerization initiator. As a result of in situ polymerization of MMA to poly-MMA in the pores of alumina, a uniform impregnation of the entire support porosity is obtained, which has been confirmed by Sorbier et al The extrudate sections were then mechanically polished under water to #4000 grit (SiC paper) using an automatic Buehler AutoMet 250 polisher. This preparation ensures sufficient resolution for spatial heterogeneity imaging.…”

“…After calibration of the two extreme porosity values with reference materials, the local porosity ε i local for each SEM image was determined by using a calibration model proposed by Sorbier et al, based on Donovan’s electron fraction mixing rule ε i normall normalo normalc normala normall = V r Z b x ( η normalb − η̅ ) V r Z b x ( η normalb − η̅ ) + V b Z r x ( η̅ − η normalr ) where V b and V r are the molar volumes of the bulk alumina phase and the PMMA resin, respectively, Z b and Z r the modified atomic numbers (Table S1, see Supporting Information), η b and η r are the backscattering coefficients of the two materials, and η̅ is the backscattering coefficient measured for the SEM image. In Donovan’s relation, the empirical exponent x of 1.4 is used, whereas Sorbier et al proposed x = 1.319 . We have used a corrected empirical exponent of x = 0.865 obtained through least-squares fitting on Monte Carlo simulation results obtained for homogeneous virtual alumina samples that lead to excellent agreement over the whole porosity range.…”

“…Furthermore, γ-alumina's spatial heterogeneity is often reported at the microscale, in the form of alumina inclusions embedded in an alumina matrix with a remarkable density difference between both phases observable on scanning electron microscopy (SEM) images. 7,10,11 These heterogeneities in the form of inclusions can influence the mechanical and transport properties of porous materials. Since alumina inclusions are easily observable by SEM, their quantification can be performed via SEM image processing methods to evaluate their impact on the mentioned properties.…”

“…Already at the nanoscale, Wang et al observed changes in local alignment of alumina nanoparticles, and, therefore, in the stacking of these nanoparticles forming the aggregates. Furthermore, γ-alumina’s spatial heterogeneity is often reported at the microscale, in the form of alumina inclusions embedded in an alumina matrix with a remarkable density difference between both phases observable on scanning electron microscopy (SEM) images. ,, …”

“…After calibration of the two extreme porosity values with reference materials, the local porosity ε i local for each SEM image was determined by using a calibration model proposed by Sorbier et al, 11…”