Cornerstone guide · Technique

SE vs BSE: Choosing a Detector on Purpose

Most operators flip between secondary and backscattered electrons by habit — SE for "normal," BSE when something looks odd. Both signals are telling you different things about the same spot, and knowing which is which changes what you conclude.

1Two signals, one beam

When the primary beam enters a sample it scatters, and two populations of electrons come back out. Secondary electrons (SE) are sample electrons kicked loose by inelastic collisions; by convention they are the ones that leave with less than 50 eV, and in practice most have only a few eV. That low energy is the whole story: they can only escape from the top few nanometers, so they report on the surface — its shape, its edges, its tilt. Backscattered electrons (BSE) are primary-beam electrons that scattered through large enough angles, often several times, to turn around and exit again. They leave with a large fraction of the beam energy, from a volume that reaches a good fraction of a micron down, and their number depends strongly on the average atomic number of what they scattered from. They report on composition — and, more subtly, on crystal orientation and on things a little below the surface.

The same pixel, the same instant, two answers. The image you get depends on which answer your detector is built to hear.

2Where each signal actually comes from

It helps to split the secondary electrons by where they were born, because the sub-populations carry different information — and this is the single most useful idea on this page.

SE1 are generated right where the focused probe enters, within a few nanometers of the beam axis. They carry the highest spatial resolution in the whole instrument; an in-lens detector at low kV on a field-emission column is essentially an SE1 microscope. SE2 are generated when backscattered electrons, on their way out, knock loose secondaries near the exit surface. They emerge from a disc roughly the size of the BSE exit area — tens to hundreds of nanometers across — and because their number tracks the BSE yield, they carry backscatter (composition) information dressed up as a secondary signal. SE3 are generated when backscattered electrons strike the pole piece, the chamber walls, or the detector housing and make secondaries there; these have nothing to do with the spot under the beam and add a diffuse, slowly varying background that is mostly noise. Some texts also name SE4, made in the column by stray beam electrons; it is small and we will leave it alone.

3The detectors, and what they really collect

The Everhart–Thornley detector (ETD). The workhorse since 1960: a scintillator behind a wire-mesh Faraday cage, coupled through a light pipe to a photomultiplier, mounted on the chamber wall off to one side. With the cage biased a few hundred volts positive it sweeps in low-energy secondaries from all over the chamber, including SE2 and SE3 and the line-of-sight BSE that happen to hit the scintillator. That indiscriminate appetite is a feature: the ETD sees around corners, so you get a well-lit, almost optical rendering of topography with soft shadows pointing away from the detector. Bias the cage negative and secondaries are rejected; what remains is the small fraction of BSE travelling straight at the detector — a high-contrast, hard-shadowed "BSE" image that is really a topographic image lit from one side. Useful for a quick look, not for composition.

In-lens (through-the-lens) SE detectors. On immersion-lens and field-emission columns, secondaries generated at the probe are drawn up the bore by the lens field and detected above the objective. This geometry strongly favors SE1 and rejects most SE3 and much SE2. The images look different — flatter, less "lit," sometimes unfamiliar — but they are the highest-resolution surface images the instrument can make, and they respond sharply to surface chemistry and charging. At low kV, this is the detector that makes sub-coating-thickness detail visible.

Solid-state BSE detectors. An annular semiconductor diode (or a segmented set of them) mounted under the pole piece, directly above the sample. Backscattered electrons with energy above a few keV generate electron–hole pairs in the diode; the large solid angle and symmetric position make the signal depend on how many BSE came out, not which direction — so the summed signal is a clean atomic-number map. Segmented designs let you subtract opposite quadrants to recover topography from BSE alone, which is how low-vacuum and variable-pressure instruments get "SE-like" images without an SE detector. Older diodes go blind below roughly 5 kV; modern ones are usable lower, but BSE imaging remains fundamentally a higher-kV game than SE.

Scintillator BSE detectors (the Robinson type and its descendants) trade some low-energy sensitivity for speed and a large collection angle; desktop instruments and low-vacuum systems often use one of these as the primary imaging detector. The same physics applies: symmetric, summed, compositional.

4Contrast: topography versus atomic number

SE contrast is mostly about tilt and edges. Secondary yield rises steeply as the local surface tilts away from the beam, because more of the generation volume lies within escape depth of the surface; the same geometry means edges, ridges, and small particles glow, since they have escape surface on several sides. Add the directional shadowing of a side-mounted ETD and you get the familiar, intuitive, "I can see the shape of that" image. The penalty is that SE yield also depends on surface chemistry, contamination, charging, and — via SE2 — composition, so brightness in an SE image is never a reliable ruler for anything.

BSE contrast is mostly about mean atomic number. The backscatter coefficient η — the fraction of beam electrons that come back out — climbs with Z, steeply among the light elements and flattening out among the heavy ones. Heavier phases are brighter, full stop, as long as the surface is flat enough that topography isn't also modulating the signal. For polished sections this is the quickest phase map there is: no etching, no stains, and a brightness scale you can interpret.

Two things the table shows at a glance. The scale is compressed at the heavy end: gold and platinum are barely distinguishable, while aluminium and silicon differ by a usable margin — so BSE separates light phases better than heavy ones. And for compounds, what matters is the mean Z weighted by composition, so a sulfide and an oxide of the same metal can sit surprisingly close together. When two phases look identical in BSE, EDS is the next question, not a longer dwell.

BSE carry a second kind of contrast that catches people out: in a crystalline sample, backscatter yield depends on how the beam aligns with the lattice, so differently oriented grains of the same phase show as subtly different grey levels. That is electron channeling contrast, the basis of ECCI, and the reason a single-phase polished alloy has visible grain structure in BSE with no etch at all. It is faint, it vanishes with surface damage, and it is a gift once you know to look for it.

5Your SE image is part backscatter

Here is the point worth taking to the bench. For a bulk sample imaged with an ETD at conventional voltages, the SE1 fraction — the part that is truly a high-resolution surface signal — is often the minority of what the detector collects. SE2 and SE3 together can outnumber it, and the ETD also catches direct BSE. So an ordinary "secondary electron" image on a multi-phase sample already contains a soft, blurred compositional layer underneath its topographic one: heavier regions are a little brighter even when they are perfectly flat. This is why an operator will sometimes say "you can see the phases in SE too" — you can, but the SE image is not telling you that; the BSE it swallowed are. It is also why in-lens images look "wrong" to people raised on the ETD: the composition wash and the side lighting are both gone, and what remains is the true surface.

The practical rule: if a brightness difference matters to your conclusion, confirm it in a detector built to measure that quantity. Flat feature, uniform tilt, brighter in BSE — that is composition. Bright in SE but not in BSE — that is topography, chemistry, or charge. Bright in both — look at both, and think.

6Choosing on purpose

The heuristics, in the order they usually apply:

• "What does it look like?" — SE, Everhart–Thornley, moderate kV (5–15), working distance around 10 mm. The intuitive image; start here.
• "What is it made of, and where?" — BSE, summed/compositional mode, 15–20 kV, short working distance so the detector sees a large solid angle. Flat or polished surfaces give clean results; rough ones mix in topography.
• "Is there fine detail on the surface?" — in-lens SE at low kV (0.5–3 kV), short working distance. Accept the unfamiliar look; it is showing you the top few nanometers and not much else.
• "It's charging and I can't coat it." — BSE, because high-energy electrons ignore the small surface potentials that scatter secondaries, and especially BSE in low-vacuum mode. You lose SE topography; gain a usable image.
• "Grains in a single-phase alloy." — BSE on a well-polished, undeformed surface, and be patient: channeling contrast is faint.
• "I'm not sure." — collect both, same field, same settings, and save both. Mixed-signal images (SE+BSE, or segmented BSE topography) are legitimate, but label them as such; the next person to read the figure deserves to know what the brightness means.

Kilovoltage deserves its own line because it changes both signals at once. Raising kV deepens the interaction volume: BSE information comes from further down (and can reveal or obscure a buried layer), SE2 spreads wider, and surface detail softens. Lowering kV collapses the volume toward the surface: SE images sharpen and become more surface-specific, BSE yield contrast holds but the diode signal weakens, and the SE/BSE distinction itself starts to blur as the beam energy approaches the energies we used to call "secondary." The conventional 50 eV cut is a convention, not physics.

7Quick reference

8Further reading

The signal physics is laid out with unusual clarity in Goldstein et al., Scanning Electron Microscopy and X-Ray Microanalysis (4th ed., Springer), whose early chapters are the source of the SE1/SE2/SE3 framing used here; Reimer's Scanning Electron Microscopy: Physics of Image Formation and Microanalysis is the deeper, more mathematical treatment. The detector itself dates to Everhart and Thornley's 1960 paper in Journal of Scientific Instruments, still a pleasure to read; Seiler's 1983 review "Secondary electron emission in the scanning electron microscope" in Journal of Applied Physics remains the standard reference on where secondaries come from. For free coursework, myScope and the University of Cambridge's DoITPoMS both have SEM modules that cover detectors and contrast well. On this site, the sample-preparation guide covers the charging problem that so often decides the SE/BSE choice for you, and the sputter-coating guide covers what a coat does to each signal.