3.4.3 Data Collection: Electron Dose Considerations
The acceleration voltage of the electron beam can be selected, and dictates the
penetrance and contrast of the electron beam, with higher energies being capable of
penetrating thicker samples due to statistically longer electron path-lengths before
being inelastically or multiply scattered. Voltage can be selected up to a given
maximum for the microscope model. Higher voltages (300 kV) are invariably better
suited to the thicker samples used in ECT and therefore the user will usually select
the highest voltage the microscope is capable of delivering; higher voltage
microscopes also tend to be better equipped for tomographic data collection. In
practical terms, 200 kV is the lowest voltage useful for tomography of samples
thicker than a couple of hundred nanometres.
Cumulative electron dose over the tilt series is a critical consideration for ECT.
Unfortunately, for every elastically scattered electron that contributes to image
formation, there are approximately three inelastically scattered electrons that
damage the specimen and contribute noise to the image [50]. This problem is
particularly acute for ECT where multiple images need to be acquired of a single
specimen, with each successive image progressively damaging the specimen. If a
high resolution subtomogram average is desired, the user must ensure that the
cumulative electron dose across the tilt series must be kept low enough to avoid
excessive damage at the molecular level, typically to <50 e
− /Å
2 levels used for
single particle analysis [51, 52]. Optimal electron dose for subtomogram averaging
can be determined by using only a subset of early frames from the tilt series to make
the final reconstruction. Here, a reconstruction based on the full tilt series can be
used for alignment, but the average can be made using only the first few—relatively
undamaged—images, leading to increased resolution [53]. This process could be
followed using different subsets of early frames to determine the optimal dose
before electron damage becomes detrimental to the average. Medium-resolution
subtomogram averages from whole cells may benefit from trading higher-resolution
information for higher contrast by increasing the electron dose to >60 e
− /Å
2 [54]
while collection of very large datasets for multiple lower-resolution averages may
benefit from reduced data quantity requirements by trading medium-resolution
information for higher contrast by using higher electron doses of greater than
120 e
− /Å
2 [35]. Finally, whole-cell tomography projects, in which noise obscures
signal below *4 nm resolution can be collected at even higher electron doses from
*120 e
− /Å
2 to *300 e
− /Å
2 , depending upon the radiation tolerance of the specimen, the principal consideration being to avoid large-scale warping or ‘bubbling’
of the specimen that will lead to poor tomographic reconstruction [55]. Because
total cumulative dose is distributed between the frames of the tilt series, but sample
thickness increases with tilt angle, exposure times (or electron flux) can be
increased as a function of tilt angle.
3 Electron Cryo-Tomography
73
The acceleration voltage of the electron beam can be selected, and dictates the
penetrance and contrast of the electron beam, with higher energies being capable of
penetrating thicker samples due to statistically longer electron path-lengths before
being inelastically or multiply scattered. Voltage can be selected up to a given
maximum for the microscope model. Higher voltages (300 kV) are invariably better
suited to the thicker samples used in ECT and therefore the user will usually select
the highest voltage the microscope is capable of delivering; higher voltage
microscopes also tend to be better equipped for tomographic data collection. In
practical terms, 200 kV is the lowest voltage useful for tomography of samples
thicker than a couple of hundred nanometres.
Cumulative electron dose over the tilt series is a critical consideration for ECT.
Unfortunately, for every elastically scattered electron that contributes to image
formation, there are approximately three inelastically scattered electrons that
damage the specimen and contribute noise to the image [50]. This problem is
particularly acute for ECT where multiple images need to be acquired of a single
specimen, with each successive image progressively damaging the specimen. If a
high resolution subtomogram average is desired, the user must ensure that the
cumulative electron dose across the tilt series must be kept low enough to avoid
excessive damage at the molecular level, typically to <50 e
− /Å
2 levels used for
single particle analysis [51, 52]. Optimal electron dose for subtomogram averaging
can be determined by using only a subset of early frames from the tilt series to make
the final reconstruction. Here, a reconstruction based on the full tilt series can be
used for alignment, but the average can be made using only the first few—relatively
undamaged—images, leading to increased resolution [53]. This process could be
followed using different subsets of early frames to determine the optimal dose
before electron damage becomes detrimental to the average. Medium-resolution
subtomogram averages from whole cells may benefit from trading higher-resolution
information for higher contrast by increasing the electron dose to >60 e
− /Å
2 [54]
while collection of very large datasets for multiple lower-resolution averages may
benefit from reduced data quantity requirements by trading medium-resolution
information for higher contrast by using higher electron doses of greater than
120 e
− /Å
2 [35]. Finally, whole-cell tomography projects, in which noise obscures
signal below *4 nm resolution can be collected at even higher electron doses from
*120 e
− /Å
2 to *300 e
− /Å
2 , depending upon the radiation tolerance of the specimen, the principal consideration being to avoid large-scale warping or ‘bubbling’
of the specimen that will lead to poor tomographic reconstruction [55]. Because
total cumulative dose is distributed between the frames of the tilt series, but sample
thickness increases with tilt angle, exposure times (or electron flux) can be
increased as a function of tilt angle.
3 Electron Cryo-Tomography
73