diffraction images collected from equally many crystals, all in different orientations [3]. However, reducing the exposure lowers the
signal, which depends on the number of scattered electrons, and it
does not necessarily change the noise. For low-dose data acquisition to be successful, noise must be controlled and the camera that
is used to record the diffraction pattern must be sensitive enough to
extract information from weak signals. To maximize useful data
obtained within the dose tolerance of the sample, it is important
to carefully plan data collection to optimally distribute the probing
electrons across the sample.
In cryo-EM, the sample is exposed to electrons that are accelerated in the column of an electron microscope. The electrons that
contribute useful information in the diffraction spots on the detector interact with the sample elastically, without loss of total kinetic
energy. A small fraction of the elastic scattering events is destructive
and can directly dislocate atoms from their chemical bonds, but the
cross-sections of these knock-on events are so low at voltages
common for cryo-EM that they have historically been ignored
[4]. For every elastic interaction, an estimated three electrons
interact inelastically [5], where the deposited kinetic energy is
ultimately transformed to heat. Most damage occurs due to electrons that lose between 5 and 100 eV during the interaction [6], by
mechanisms that are assumed to be similar to those that cause
damage under X-ray radiation. The deposited energy predominately excites or ionizes valence electrons, which breaks chemical
bonds and produces electrons and free radicals. The electrons and
their associated Auger electrons liberated during these primary
damage events are mobile even at 77 K and can continue to break
bonds, even at cryogenic temperatures [7]. Free radicals, on the
other hand, generate cascades of secondary chemical reactions [8],
and spread through the crystal by thermal diffusion; the damage
caused by these events can be controlled by lowering the datacollection temperature. The absorbed energy, which is related to
radiation damage, depends on the chemical composition of the
crystal and its surrounding mother liquor as well as the energy of
the incident electrons. For instance, the two disulfide bonds in
proteinase K have been observed to exhibit breakage at different
rates, even though all data collection parameters are identical
[9, 10].
The individual molecules in the crystal lattice are affected as
soon as the first electrons enter the sample, but since every atom in
the crystal contributes to every reflection on the diffraction pattern,
these effects may not be immediately apparent during data collection. In contrast, the gradual deterioration of the lattice is readily
apparent in the diffraction patterns: as the accumulated exposure
increases, fine lattice features are disrupted, and spots become
fainter, beginning with the weak spots at high resolution. As exposure continues, such global radiation damage causes spots at lower
310
Johan Hattne
signal, which depends on the number of scattered electrons, and it
does not necessarily change the noise. For low-dose data acquisition to be successful, noise must be controlled and the camera that
is used to record the diffraction pattern must be sensitive enough to
extract information from weak signals. To maximize useful data
obtained within the dose tolerance of the sample, it is important
to carefully plan data collection to optimally distribute the probing
electrons across the sample.
In cryo-EM, the sample is exposed to electrons that are accelerated in the column of an electron microscope. The electrons that
contribute useful information in the diffraction spots on the detector interact with the sample elastically, without loss of total kinetic
energy. A small fraction of the elastic scattering events is destructive
and can directly dislocate atoms from their chemical bonds, but the
cross-sections of these knock-on events are so low at voltages
common for cryo-EM that they have historically been ignored
[4]. For every elastic interaction, an estimated three electrons
interact inelastically [5], where the deposited kinetic energy is
ultimately transformed to heat. Most damage occurs due to electrons that lose between 5 and 100 eV during the interaction [6], by
mechanisms that are assumed to be similar to those that cause
damage under X-ray radiation. The deposited energy predominately excites or ionizes valence electrons, which breaks chemical
bonds and produces electrons and free radicals. The electrons and
their associated Auger electrons liberated during these primary
damage events are mobile even at 77 K and can continue to break
bonds, even at cryogenic temperatures [7]. Free radicals, on the
other hand, generate cascades of secondary chemical reactions [8],
and spread through the crystal by thermal diffusion; the damage
caused by these events can be controlled by lowering the datacollection temperature. The absorbed energy, which is related to
radiation damage, depends on the chemical composition of the
crystal and its surrounding mother liquor as well as the energy of
the incident electrons. For instance, the two disulfide bonds in
proteinase K have been observed to exhibit breakage at different
rates, even though all data collection parameters are identical
[9, 10].
The individual molecules in the crystal lattice are affected as
soon as the first electrons enter the sample, but since every atom in
the crystal contributes to every reflection on the diffraction pattern,
these effects may not be immediately apparent during data collection. In contrast, the gradual deterioration of the lattice is readily
apparent in the diffraction patterns: as the accumulated exposure
increases, fine lattice features are disrupted, and spots become
fainter, beginning with the weak spots at high resolution. As exposure continues, such global radiation damage causes spots at lower
310
Johan Hattne