The SEM image is effectively made up of lines of image points, each point being the size of the beam spot at the sample surface. The ability of the SEM to resolve fine structures is limited by the diameter of this spot size (probe size). It is also limited by the number of electrons contained within the probe. If the probe is too small in relation to the area being imaged, it spends too little time on each image point to provide sufficient signal to form a good quality image. There is a finite relationship between magnification and the optimum probe size and it does vary from specimen to specimen.

Each dwell time (seen as a spot in the image) generates electrons that are used to make up the image on the screen. We see edges and dips and bumps on a sample because of changes in the number of electrons coming off the sample at that point. As we go up in magnification, and drop our beam probe size down to a smaller and smaller spot, we see more detail (see line 2 in the diagram). But there is a limit. The limit of magnification is the point where no variation in signal (electrons generated from the sample) is obtained from adjacent points on the sample. This performance limit is dependent on the composition and structure of the specimen being examined. For example, specimens such as metals with a high atomic number (Z) produce a high yield of electrons and achieve a higher useful magnification than low Z samples (e.g. carbon and plastic).

Secondary electron (SE) images

For routine scanning electron microscope images, secondary electrons (SE) are usually used to image the surface. Secondary electrons are low energy electrons formed by inelastic scattering and have energy of less than 50eV. The low energy of these electrons allows them to be collected easily. This is achieved by placing a positively biased grill on the front of the SE detector, which is positioned off to one side of the specimen. The positive grill attracts the negative electrons and they go through it into the detector. This is the case for the Everhart-Thornley detector, which is most commonly used but some machines have another kind of in-lens SE detector.

The major influence on SE signal generation is the shape (topography) of the specimen surface. Secondary electrons provide particularly good edge detail. Edges (and often pointy parts) look brighter than the rest of the image because they produce more electrons.
The image below shows protuberances (bumps) on the wing of an insect. Notice the whiter edge to each bump.

To increase the yield of SE emitted from the specimen, heavy metals such as gold or platinum are routinely used to coat specimens. An extremely thin layer is applied (~10 nm). This coating is applied for two main reasons:
1. Non-conductive specimens are often coated to reduce surface charging that can block the path of SE and cause distortion of signal level and image form.
2. Low atomic number (Z) specimens (e.g. biological samples) are coated to provide a surface layer that produces a higher SE yield than the specimen material.

Because secondary electrons have very low energies, only those produced at the surface of the sample are able to escape and be collected by the SE detector. Electrons emitted from a surface that faces away from the detector or which is blocked by the topography of the specimen, will appear darker than surfaces that face towards the detector. This topographical contrast due to the position of the SE detector is a major factor in the 'life-like' appearances of SE images.
In the image of the silicon wafer below, the copper balls are brightest towards the bottom right, indicating that the detector is in the bottom right corner. It is, however, not the only factor that contributes to the contrast and brightness in an SEM.

The contribution of BSE to images collected with the SE detector

The primary function of the SE detector is to attract low energy secondary electrons. These SEs are generated from approximately the top 15nm of the surface. Unless the SEM is specially set up to minimise the BSE contribution, the image produced by the detector will always contain an amount of sub-surface information derived from high energy BSE. As a general rule, the higher the kV the more sub-surface information is picked up by the detector due to various backscattered effects (elastic scattering effects).

At 2kV you will see a lot more surface detail than at 20kV, but this surface detail may be due to contamination. One important skill in operating an SEM is to choose the correct kV for your specimen such that you gather information from the depth of the specimen that interests you, with the least contribution from surface contamination above or unimportant structures below.
The following image is from the secondary electron detector and shows a silicon wafer surface at the same magnification but at different beam energies (kV): a = 5kV, b = 10kV, c = 15kV, d = 20kV. Note that the features in image d are much more rounded than in image a because there is more information coming from below the surface of the sample.

So far only the secondary electrons produced by interaction of the primary electron beam with the sample have been discussed. These are termed SE1 electrons. There are a number of different types of secondary electrons. Backscattered electrons can generate secondary electrons. These are termed 'SE2'. Interaction of the beam with the sample chamber, pole piece etc. can also produce secondary electrons. These are termed 'SE3'.