Scanning 1938–2002

Interactive exhibit · the scanning transmission electron microscope

From the first scan to single atoms

A scanning microscope builds its picture one point at a time with a focused probe. Make that probe smaller and brighter and you see more. Drive three machines that did it: von Ardenne’s in 1938, Crewe’s in 1970 and the corrected VG HB501 in 2002.

Drag the image to move the specimen · scroll it to change the magnification

    What limits the probe

    —probe size (FWHM of the core)

    Micrographs

    No recordings yet. Press Record to keep the next full frame.

    How it works

    A scanning transmission microscope focuses the beam into a probe and moves it across a thin specimen; detectors below (or, in the HB501, above) record what comes through at each point. Everything you see is computed from these rules each frame.

    Brightness

    Current costs size

    A source of brightness B puts at most I = B · (π d²/4) · (π α²) into a probe of size d and semi-angle α. Halve the probe and the current falls by four. A hot tungsten filament (~10⁴ A/cm²·sr) cannot fill a nanometre probe; a cold field emitter (~10⁸–10⁹) can.

    Diffraction

    A wider cone, a smaller spot

    The probe can be no smaller than the diffraction disc of its aperture, about 0.5 λ/α. At 120 kV, λ = 3.35 pm, so a 30 mrad cone allows a spot of about 0.6 Å.

    Aberrations

    The flat region

    Rays at larger angles are focused too strongly by spherical aberration. Only within the angle where the phase error stays below π/4 do the waves add up into a sharp core; the rest of the current spreads into a halo. Defocus and astigmatism shrink that angle; the corrector widens it.

    Z-contrast

    Heavy atoms shine

    Electrons that pass close to a nucleus are scattered to high angles, roughly as Z^1.7. An annular detector that collects only those makes a uranium atom (Z = 92) about a hundred times brighter than a carbon atom (Z = 6).

    Ronchigram

    Seeing the aberrations

    With the probe focused near a thin amorphous film and the scan stopped, the detector plane shows a shadow image. Where the aberration phase is flat, the film is magnified so much that it looks smooth. Tuning a corrector means making that smooth centre as wide as possible.

    Scanning

    Time is part of the image

    Each pixel is recorded at a different moment. Slow scans give more electrons per pixel, so less noise, but any drift during the frame shears and smears the picture. In 1938, one frame took many minutes on a film drum.

    Three probes, three limits

    Each instrument removed the limit of the one before: von Ardenne showed that scanning works but had too dim a source; Crewe’s field-emission gun made the probe bright enough for single atoms; the aberration corrector made it smaller than an atom.

    There was no screen to watch in 1938: the probe crossed the specimen while photographic film on a drum turned in step with it, and the picture appeared only when the film was developed.

    1. 1935Max Knoll scans an electron beam across a specimen to form an image: the principle of the scanning electron microscope.
    2. 1938Manfred von Ardenne builds a scanning microscope working in transmission, the first STEM, with a demagnified probe and film recording on a drum.
    3. Late 1960sAlbert Crewe’s group at the University of Chicago develops a practical cold field-emission gun.
    4. 1970Crewe, Wall and Langmore report images of single uranium and thorium atoms on thin carbon films.
    5. 1970sVacuum Generators (VG) builds dedicated STEMs, with the gun at the bottom and the beam going up.
    6. 1997–98Working aberration correctors are demonstrated: Krivanek and Dellby for STEM in Cambridge, Haider, Rose and co-workers for TEM.
    7. 2002Batson, Dellby and Krivanek report a probe smaller than one ångström on IBM’s corrected VG HB501 and resolve silicon dumbbells 0.78 Å apart.
    8. 2010Corrected STEMs identify single atoms by both their scattering and their energy-loss spectra, atom by atom.

    Your recorded frame