Comparison of phase contrast transmission electron microscopy with optimized scanning transmission annular dark field imaging for protein imaging

Henderson has already shown that electron microscopy should be superior to X-ray and neutron diffraction for determining protein structure with minimum radiation damage. Since the contrast for a molecule embedded in vitreous ice is very low, it is conceivable that dark field imaging would be superior to bright field phase contrast microscopy. A detailed analysis of contrast and signal/noise for both imaging modes is presented. Annular dark field scanning transmission microscopy gives improved contrast and equivalent signal/noise to phase contrast TEM when the molecule is the same thickness as a vitreous ice embedding medium. For a constant embedding medium thickness of 200 إ the contrast is equivalent to phase contrast TEM but the signal/noise is 5 times worse. Even with an efficient detector that only excludes scattering less than 5 mrad there is insufficient signal at a dose of 5 electrons/إ2 to produce an image with more than 1 electron/per pixel. For larger molecules (>100 إ thick which corresponds to 420 kDa for spherical molecules) the weak phase object approximation used to analyse a phase contrast image no longer applies at 100 kV. This limit could be extended to about 200 إ (about 3 MDa) if a 400 kV microscope were used.