SEM Image Analysis in the Laboratory: Particle Sizing

# SEM Image Analysis in the Laboratory: Particle Sizing

SEM image analysis is a direct, image-based method for measuring particle size. It resolves individual particles and records their true two-dimensional morphology, which suits irregular, agglomerated, or mixed powders. The workflow covers sample preparation, image acquisition, threshold-based segmentation, and statistical conversion into a size distribution. It is applied to ceramics, metal powders, pharmaceutical excipients, pigments, and wear debris.

Principle & mechanism

An SEM scans a focused electron beam across the specimen in a raster pattern. It collects secondary or backscattered electrons to form a magnified image. Accelerated electrons have a far shorter wavelength than visible light, so the effective resolution reaches the nanometer scale. Particles from several nanometers to tens of micrometers can be distinguished clearly. Particle sizing itself is performed on the digital image, not inside the microscope. Calibration bars embedded by the instrument define the pixel-to-length scale at each magnification. Software then detects particle boundaries by grayscale thresholding or edge detection. It measures projected features such as equivalent circular diameter, Feret diameter, and aspect ratio, and aggregates thousands of measurements into a statistically valid size distribution.

Sample Types and Preparation

The method applies to dry powders, suspensions after drying, metallographic cross-sections, filter deposits, and wear debris recovered from lubricants. Conductive materials such as metal powders can often be imaged after simple dispersion on carbon adhesive tabs. Non-conductive ceramics, polymers, and biological particles require a thin conductive coating, typically sputtered gold or platinum. This suppresses charging artifacts that distort particle outlines. Dispersion is the critical preparation step. Agglomerates must be broken up gently, for example in a suitable liquid with ultrasonic agitation followed by drop-casting onto a substrate. This ensures the measured objects are primary particles rather than clusters. Cross-sectional samples are mounted, ground, and polished so that sectioned particle contours remain sharp. Every submission should carry careful labeling and chain-of-custody records.

Testing and Measurement Workflow

A typical workflow proceeds in five steps. First, the operator verifies instrument calibration against a certified magnification standard. They record accelerating voltage, working distance, and detector settings. Second, particles are dispersed at low areal density, ideally with minimal overlap, because touching particles bias results toward larger sizes. Third, images are captured at magnifications chosen so that the smallest particles span a sufficient number of pixels. Multiple magnifications may be combined to cover a broad size range. Fourth, representative fields are selected without bias toward large or visually striking particles, and stage positions are randomized where practical. Fifth, images are exported in an unprocessed format together with the pixel scale for analysis. For failure-related projects, the same session often feeds compositional data into Connector and Relay Failure Analysis Testing or bearing investigations.

Image

Processing and Particle Size Distribution
Processing converts grayscale images into measurable binary objects. After thresholding, morphological operations such as watershed splitting separate lightly touching particles. Manual review then rejects artifacts, coating debris, and edge-truncated particles. For each accepted particle, the software computes the equivalent circular diameter, the diameter of a circle of equal projected area, together with Feret minimum and maximum lengths. Results are reported as number-based distributions, or converted to volume-based distributions when density is assumed uniform. Common descriptors include D10, D50, D90, the arithmetic or geometric mean, span, and the histogram on linear or logarithmic axes. Sample statistics matter. Several hundred to several thousand particles are normally counted, and the required count grows when the distribution is broad. Analysts should state the counting basis clearly, since number and volume distributions of the same sample can look very different.

Accuracy, Repeatability, and Resolution

Measurement uncertainty comes from several controllable sources. Pixel calibration error and the edge-definition rule applied during thresholding each shift diameters slightly. Magnification should therefore be selected so the smallest particle of interest spans at least roughly ten pixels. Sample preparation dominates repeatability. A second dispersion of the same powder can shift the reported D50 if the agglomeration state changes. Operator segmentation choices introduce subjectivity, which modern automated and machine-learning-assisted segmentation reduces but does not eliminate. Practical ways to demonstrate repeatability include round-robin style internal replicate measurements, documented threshold rules, and blind re-measurement of archived images. Relative to laser diffraction, SEM image analysis counts far fewer particles, so distribution tails are less statistically robust. However, individual-particle accuracy and morphological detail are far higher. Both techniques are often run together and cross-checked.

Applications and Reporting Standards

Applications span quality control of metal and ceramic powders, additive-manufacturing feedstock qualification, pharmaceutical particle characterization, pigment and filler inspection, and debris analysis in machinery health monitoring programs such as Bearing testing Services. The method also supplies supporting morphology evidence in packaging defect investigations alongside Chemical Analysis for Flexible Packaging. A compliant report states the instrument type, accelerating voltage, coating material, calibration verification, number of particles counted, measurement basis, distribution table with D10/D50/D90, representative images at stated magnification, and the segmentation rules applied. ISO 13322-1, which governs image-analysis-based particle size measurement, is the recognized methodological reference. Reported results should carry a statement of uncertainty or repeatability. Raw image sets should be archived so that any reported distribution can be independently reprocessed and verified.