Four detectors, same electrons, wildly different lore.
In scanning TEM, a focused electron probe sweeps across the sample point by point, and detectors below collect whatever scatters. The defining difference between one detector and another is the range of scattering angles it catches. Yet that single choice decides whether heavy atoms glow, light atoms appear, or contrast flips entirely.
When the probe strikes the sample, most electrons pass nearly straight through; some are deflected a little; a few, those that pass close to a heavy nucleus, are kicked out to large angles, Rutherford-style. Below the sample sits a stack of detectors: a BF (bright-field) disc on the axis, and concentric rings, ABF (annular bright-field), ADF (annular dark-field), and HAADF (high-angle annular dark-field), catching progressively larger angles.
On most instruments you don't move the detectors to change the angular ranges; you change the camera length: an effective lens zoom that spreads or compresses the scattering pattern across the fixed rings.
Because each detector integrates a different slice of scattering, the same scan yields different images simultaneously. The high-angle signal grows steeply with atomic number (roughly Z1.7), so HAADF is "Z-contrast": heavy columns bright, background dark, and, in thin specimens, far fewer reversals to trip over (thickness and channeling can still modulate it). ABF keeps a partial bright-field character that makes light columns, oxygen, nitrogen, lithium, visible as dark spots; under specially favourable conditions it has even revealed hydrogen. BF-STEM behaves like a tiny conventional TEM image, interference quirks included.
Typical working ranges (they shift with camera length, and conventions vary): BF collects roughly 0–10 mrad on axis; ABF a ring within the bright-field disc at about 10–25 mrad; ADF (sometimes "LAADF") around 25–60 mrad, where diffraction and strain still contribute; and HAADF beyond ~60 mrad out to 200 mrad or more, where scattering is mostly thermal-diffuse and beautifully incoherent. For reference, the probe itself converges at ~10 mrad on an uncorrected instrument like a Talos 200i, and ~30 mrad on a probe-corrected Spectra 300: the convergence angle sets where "bright field" ends.
High-angle scattering comes from close encounters with the nucleus (screened Rutherford scattering with cross-section rising as ~Z1.7–2) and is dominated by thermal diffuse scattering. Phases scramble, interference washes out, and each atom contributes intensity independently: the image is (to a good approximation) a simple convolution of the probe with an object function. That's why HAADF images are far more directly interpretable than HRTEM images, which need simulation.
The principle of reciprocity says a BF-STEM image with a small collector is equivalent to a conventional TEM image with the beam path reversed: source ↔ detector. Everything you know about CTFs and defocus in TEM carries over to BF-STEM, which is exactly why BF-STEM shows phase-contrast quirks that HAADF doesn't.
On a probe-corrected instrument, correcting spherical aberration lets you open the probe-forming aperture to ~30 mrad, shrinking the probe below 1 Å: sub-Ångström Z-contrast imaging is the reward.
@misc{tripathy2026stemdetectorsbfabfadfhaa,
author = {Tripathy, Manisha},
title = {STEM Detectors: BF, ABF, ADF & HAADF},
year = {2026},
howpublished = {\url{https://untetheredatom.com/tem/stem-detectors-guide}},
note = {Interactive teaching resource}
}