Process-microstructure mapping in additively manufactured components from SEM images

Pratyush Kumar, Texas A & M University, Mechanical Engineering

Initial Publication Date: October 7, 2025

Summary

This activity introduces students to the relationship between manufacturing process parameters and resulting microstructural features by analyzing Scanning Electron Microscopy (SEM) images. Students use MATLAB to quantify image-based descriptors such as grain size, porosity, texture, and morphological features, besides the compressed features from the Autoencoder. They then correlate these descriptors with processing parameters (e.g., laser power, scan speed, printing strategy in additive manufacturing) to develop insight into how process choices govern material microstructure and, ultimately, performance.

Engineering | Graduate/Professional | Designed for In-Person

Expand for more detail

Share your modifications and improvements to this activity through the Community Contribution Tool »

Learning Goals

Conceptual:

Understand how process parameters (e.g., laser power, scan speed) influence microstructural features (grain morphology, porosity, texture).

How to build Autoencoder for extracting compressed features.

Connect image-based observations to structure–property relationships.

Computational (MATLAB):

Import and preprocess SEM images.

Compute microstructure descriptors:

Porosity fraction via thresholding.

Grain size distribution via segmentation.

Texture or anisotropy.

Morphological metrics.

Correlate descriptors with process parameters using regression/plots.

Higher-Order Skills:

Critical thinking about experimental variability.

Data-driven modeling of materials processing.

Context for Use

Appropriate for upper-level undergraduate or graduate courses in Materials Science, Additive Manufacturing, and Computational Mechanics.

Works for class sizes 10–30 (lab groups) or larger if SEM images are available in advance.

Students should have prior exposure to basic materials processing concepts and microstructure interpretation.

Familiarity with MATLAB image processing functions is helpful but not necessary.

Description and Teaching Materials

Activity Description

Data Introduction: Students are given a dataset of SEM images of samples fabricated with different process parameters (e.g., laser power levels).

A CSV file links each image ID to its processing parameters.

Image descriptor extraction is done using MATLAB.

Students implement MATLAB scripts to compute features: mean grain area, porosity, pore eccentricity, FFT-based texture descriptors, and boundary length density.