For biological samples the effective atomic number is low, therefore inelastic
scattering events are more probable than elastic events (2.4). Since inelastic scattering generally involves the ionization of atoms, and since this often destroys the
local atomic configuration, the relation between the inelastic and elastic scattering
cross-section immediately reveals that specimens with light elements are more
likely to suffer from radiolysis. This radiation damage is a significant impediment to
electron microscopy of biological specimens.
The inverse of the cross section r el multiplied by the number density of atoms
per unit volume n a is interpreted as the mean distance or mean free path (MFP)
between elastic collisions. The MFP is an excellent guideline for understanding the
limit of sample thickness for electron microscopy imaging.
The MFP for inelastic scattering by vitreous ice has been estimated to be 200 nm
for 120 kV electrons [3, 23]. The ratio of MFP for elastic to inelastic scattering
depends approximately as 20/Z, where Z is the atomic number. Thus for biological
material dominated by the light elements carbon, nitrogen, and oxygen (Z = 6, 7,
8), for an electron traversing one elastic MFP, there would be three inelastically
scattered electrons (2.4). The inelastic MFP also sets a limit for optimal TEM
tomography conditions.
3 This is because for TEM tomographic imaging of biological specimens, the interpretable signal originates in the elastic scattering. The
inelastically scattered electrons contribute an unfocused haze and so should be
blocked in an energy filter that sits before the camera. For thicknesses equal to one
inelastic MFP, the signal is reduced by 1/e. When tilted to 60 degrees the projected
thickness doubles, so only a fraction 1/e
2 of the signal remains. For STEM imaging,
on the other hand, the practical thickness limit is the elastic MFP, because there is a
deleterious effect on the signal from multiple elastic scattering. This still allows for
imaging by STEM tomography of specimens that are at least three times thicker
than by TEM. With some compromise in resolution the thickness range can be
extended further.
2.2.2 STEM Contrast of Biological Samples
As described in the introduction, a STEM image is formed pixel by pixel by
scanning a focused electron probe over the sample; transmitted electrons are
counted at each point by a series of detectors. Most commonly used are annular DF
(ADF) and high-angle annular DF (HAADF) detectors to collect electrons that are
scattered out of the primary electron beam (Fig. 2.1). These detectors record
3
It should be stressed that we consider here thick samples for which STEM imaging offers
advantages over TEM. For the opposite extreme, that is for thin unstained samples up to a few
100s of nanometers, TEM proves superior. For such samples that are considerably thinner than the
inelastic MFP, and for which the weak phase approximation holds, TEM phase contrast offers
superior contrast and signal-to-noise-ratio compared with the STEM dark field [24].
38
S. G. Wolf et al.
scattering events are more probable than elastic events (2.4). Since inelastic scattering generally involves the ionization of atoms, and since this often destroys the
local atomic configuration, the relation between the inelastic and elastic scattering
cross-section immediately reveals that specimens with light elements are more
likely to suffer from radiolysis. This radiation damage is a significant impediment to
electron microscopy of biological specimens.
The inverse of the cross section r el multiplied by the number density of atoms
per unit volume n a is interpreted as the mean distance or mean free path (MFP)
between elastic collisions. The MFP is an excellent guideline for understanding the
limit of sample thickness for electron microscopy imaging.
The MFP for inelastic scattering by vitreous ice has been estimated to be 200 nm
for 120 kV electrons [3, 23]. The ratio of MFP for elastic to inelastic scattering
depends approximately as 20/Z, where Z is the atomic number. Thus for biological
material dominated by the light elements carbon, nitrogen, and oxygen (Z = 6, 7,
8), for an electron traversing one elastic MFP, there would be three inelastically
scattered electrons (2.4). The inelastic MFP also sets a limit for optimal TEM
tomography conditions.
3 This is because for TEM tomographic imaging of biological specimens, the interpretable signal originates in the elastic scattering. The
inelastically scattered electrons contribute an unfocused haze and so should be
blocked in an energy filter that sits before the camera. For thicknesses equal to one
inelastic MFP, the signal is reduced by 1/e. When tilted to 60 degrees the projected
thickness doubles, so only a fraction 1/e
2 of the signal remains. For STEM imaging,
on the other hand, the practical thickness limit is the elastic MFP, because there is a
deleterious effect on the signal from multiple elastic scattering. This still allows for
imaging by STEM tomography of specimens that are at least three times thicker
than by TEM. With some compromise in resolution the thickness range can be
extended further.
2.2.2 STEM Contrast of Biological Samples
As described in the introduction, a STEM image is formed pixel by pixel by
scanning a focused electron probe over the sample; transmitted electrons are
counted at each point by a series of detectors. Most commonly used are annular DF
(ADF) and high-angle annular DF (HAADF) detectors to collect electrons that are
scattered out of the primary electron beam (Fig. 2.1). These detectors record
3
It should be stressed that we consider here thick samples for which STEM imaging offers
advantages over TEM. For the opposite extreme, that is for thin unstained samples up to a few
100s of nanometers, TEM proves superior. For such samples that are considerably thinner than the
inelastic MFP, and for which the weak phase approximation holds, TEM phase contrast offers
superior contrast and signal-to-noise-ratio compared with the STEM dark field [24].
38
S. G. Wolf et al.