Microstructure Analysis: Metallography Steps and How to Read Steel Phases

It is the steel's internal structure at high magnification (tens to thousands of times) after cutting, polishing and etching to reveal the phases.

Common phases include martensite (hard), pearlite, ferrite, plus grain—together they show whether heat treatment succeeded.

Choose a location and orientation that answer the question: through a thread root for case depth or normal to a crack. Mount, grind, fine-polish and finish without rounding the edge or overheating the specimen.

The unetched surface can reveal some inclusions, pores and cracks. Appropriate etching reveals phases and grain boundaries. ISO/TR 20580:2022 and ASTM E3/E407 are references for preparation and etching practice.

Poor preparation creates scratches, pull-out and smeared layers that can be mistaken for real defects, so record the section, preparation and etchant.

• Distinguishes ferrite, pearlite, bainite and martensite to verify the heat-treatment response.
• Measures surface case depth and reveals any decarburization layer.
• Checks grain size, cracks, porosity or inclusions.
• Compares passing and failed parts from the same lot.
• Supports failure analysis when combined with history, hardness, chemistry and fracture evidence.

Appearance changes with alloy, etchant, magnification, lighting and preparation. Do not identify a phase from “black versus white” alone; compare material, process history and a suitable reference.

Low-carbon steel before hardening often shows ferrite and pearlite; a suitable quench can create martensitic structure; tempering changes its fine appearance with time and temperature. Companion hardness data checks whether the interpretation matches properties.

For high-confidence work, new lots, or when a quality issue is suspected.

For acceptance inspection, define section, sample count, magnification, criterion and reference. For failure work, preserve passing and failed samples from the same lot with process records.

A useful report includes images with scale bars, section location, etched/unetched condition, magnification, observations and limitations—not an image without context.

Metallography is destructive and samples a small area. One normal field does not prove an entire lot is normal, while one abnormal field must be judged as representative or local.

Root-cause conclusions should combine hardness profile, material certificate/chemistry, furnace records, coating history and fracture evidence as the problem requires.