In this method, a focused electron beam (5 kV to 50 kV) is scanned over the sample in parallel lines. The particles are fixed onto a planar substrate, and are normally coated with a thin conductive layer, often an amalgam of gold and palladium. The electrons interact with the sample, producing an array of secondary effects, such as back-scattering, that can be detected and converted into an image. The image can then be digitized and presented to an image analyzer, which uses complex algorithms to identify individual particles and record detailed information about their morphology. In this manner, size and shape can be accurately assessed, but only for a relatively small population of particles (a few hundred at best). By comparison, most ensemble techniques (e.g., LAS) sample thousands or tens of thousands of particles simultaneously. Scanning electron microscopy (SEM) is therefore a particle counting technique and produces a number-weighted size distribution. In actuality, SEM measures the projected surface area diameter or length.
Limitations and errors in the SEM method generally arise from two sources: sample preparation and image distortions or irregularities. Probably the single largest source of error is sample preparation. In order for the analyzer to avoid confusing single particles that are touching (coincident) with larger particles or agglomerates, the primary particles must be well dispersed in a monolayer on the substrate surface, and must be clearly separated from each other. Specimens formed by drying of a dilute suspension onto a SEM support often contain agglomerates that form during the drying process. Specimens formed from a dry powder always exhibit agglomerative artifacts, and are not suitable for size assessment. In addition, the particles in the test specimen must be homogeneously distributed over the measurement field, so that the analysis accurately reflects the true distribution. Segregation of particles during sample preparation can lead to heterogeneous deposition, which can cause large systematic errors. An additional source of uncertainty comes from the poor statistical sampling that results from the relatively low number of particles counted by SEM analysis, compared with ensemble methods like LAS or other single particle counting techniques like EZS.
A practical resolution of 15 nm to 20 nm can be expected for SEM, establishing a lower limit for accurate particle size analysis of about 0.2 µm in secondary electron mode. An order of magnitude improvement can be obtained using the newer field emission SEM (FESEM), which has a practical resolution of about 1 nm. The FESEM uses lower electron voltages and generally does not require a conductive coating. The upper limit for imaging techniques like SEM is statistically-limited, since fewer large particles will be present in the same imaging field, and thus as the particle size increases it will require a prohibitively large number of separate specimens to obtain a statistically acceptable number of particles. This counting issue will be especially important at the upper size range for cement, where particles approach 100 µm in diameter. The image analyzer can be calibrated using a standardized graticule.
The primary benefit of SEM analysis is that it provides highly detailed information about not only particle size, but also particle shape, surface texture and chemical composition, and at resolutions not approachable by other techniques. Transmission electron microscopy (TEM), which measures the transmitted electron beam after it passes through the sample, is applicable for particle sizing in the extreme lower size limit, below 0.2 µm, although much of the three dimensional information is lost in this case. Drawbacks of using electron microscopy as a routine sizing method are time, high cost and the high level of operator expertise and training necessary.