When examining any specimen using the SEM – Scanning Electron Microscope – provided that it is equipped to do X-ray microanalysis and digital image analysis, we generally have a choice of three basic activities. We can image the specimen, we can analyse it using X-ray microanalysis and we can measure it using digital image analysis.
Most SEM imaging is either secondary electron imaging (can be shortened to se, sei) or backscattered electron imaging (bs, bse, be, bei). There are other, less common, types we won’t cover here, such as cathodoluminescence or specimen current imaging.
Secondary electrons are low energy electrons generated within the specimen, typically with energies of just a few electron volts (eV). Because the secondary electrons are of low energy, only those generated very near the surface of the specimen (maybe a few nanometres) can escape and be detected; consequently, they convey a lot of information about the specimen surface. In general, secondary electron imaging shows most detail when viewing a fracture surface. A pure secondary electron image of a flat surface won’t show any detail at all. In practice, the secondary electron detector is also sensitive to backscattered electrons, so the se image is likely to contain a small backscatter signal.
Backscattered electrons are primary electrons that are scattered by the specimen through more than 90 degrees and re-emerge from the specimen surface. Primary electrons are the electrons that travel down the microscope column and scan the specimen surface – ie: the electron beam. They are much more energetic than secondary electrons, with a maximum energy dependent on the accelerating voltage.
Backscattered electron imaging shows atomic number contrast in the specimen because the proportion of primary electrons that are backscattered is strongly dependent on the atomic number of the specimen. This proportion is termed the backscatter coefficient.
Most specimens contain compounds, rather than pure elements, and the backscatter coefficient of each compound is then dependent on its mean atomic number. For example, in cement terms this means that calcium oxide is brighter than alite (alite is impure tricalcium silicate) and alite is brighter than belite (belite is impure dicalcium silicate) or silica or limestone.
These differences in brightness, or grey level, form the backscattered electron image. Provided that the specimen is flat (typically, this would be a polished section) the resulting image is essentially an atomic number map of the specimen. The great strength of bs imaging is not that it reveals fine detail – it doesn’t when compared with se – but that it usually differentiates clearly between the different minerals present.
Because backscattered electrons have much higher energies compared with secondary electrons, backscattered electrons do not convey as much detail of surface topography – they can travel several microns through the specimen before re-emerging at the surface. While fracture surfaces, or any rough surface such as dust on tape, can be imaged using either se or bs, more surface detail will be seen if se imaging is used, while compositional information will be given in the bs image.
In summary, to image a fracture surface it will probably be best to use secondary electron imaging. Image resolution will improve with a smaller spot size and a shorter working distance. Also, the lower the SEM accelerating voltage the better the image resolution, because it reduces the range over which re-emerging backscattered electrons generate secondary electrons. Very often, an accelerating voltage is used for secondary electron imaging that is much too high, sometimes 20 kV or even 30 kV. To find the optimum accelerating voltage it will be necessary to experiment, but for a typical SEM, an accelerating voltage of 5 kV-10 kV will probably be about right. Some SEMs are designed to produce se images at very low accelerating voltages, eg: 1 kV-5 kV, or even less.
For backscattered electron imaging, again a smaller spot size and shorter working distance will be beneficial for resolution. However, a larger spot size will probably required for bs imaging than for se imaging, because there are proportionally fewer backscattered electrons and the detector is generally less-efficient at collecting them. A higher accelerating voltage will also probably be needed, perhaps 20 kV. Again, finding the optimum conditions is a matter of experimentation.
Chapters 4 and 5 of “Scanning Electron Microscopy of Cement and Concrete” have a lot more information on sem cement imaging as well as SEM terminology and identification of different materials.