anisotropic absorption, radiation damage, beam intensity variations, and temporal non-uniformity in the response of the detector
all cause intensities to exhibit non-random variations over the
course of the measurement [35]. Ideally, all these effects are
accounted for during scaling, before intensities integrated from
the first observations are merged with those measured during
later stages. However, scaling a multicrystal dataset assumes that
all individual data originate from identical crystals, which is never
the case in practice. The effect of lattice discrepancies on the scaled
and merged intensities depends on the degree of non-isomorphism.
Multicrystal merging presents an opportunity to obtain a minimally damaged dataset, as frames recorded from a highly exposed
crystal can be discarded in favor of the first frames recorded from a
different sample. If enough datasets are available, merging a highmultiplicity dataset from short exposures can thus reduce errors and
lead to overall better results [36]. Even though MicroED often
requires data from a 10
–20
wedge for accurate indexing and
integration, not all frames need to be merged into the final dataset.
A good strategy for multicrystal data collection is often to alternate
between collecting data starting at high and low tilt. In cases of
preferred crystal orientation, this will increase the probability that
reciprocal space is evenly exposed, and that damage is not concentrated to specific wedges of reciprocal space. In the limit, only the
first frame from each dataset would be merged, which requires data
from very many crystals. Increasing automation now has the potential to make such a strategy feasible [37].
4 Camera Considerations
The quality of the recorded diffraction images is paramount in
MicroED, as they constitute the only experimental measurement
on which the final atomic model is based. Weak exposures generally
yield noisy images that cannot be accurately integrated, whereas
high exposures may fall outside the linear range of the camera. An
ideal camera for diffraction measurement would account for every
electron that interacts with the sample; it must have high dynamic
range in order to accurately record both the intense, low-resolution
reflections as well as the faint, high-resolution spots on the same
image.
The ability to accurately detect very low electron counts is
critical when the exposure is attenuated in the interest of reducing
radiation damage. A particularly attractive camera option is
provided by electron-counting, which owing to the absence of
read-out noise promises to accurately record weak, high-resolution
reflections. However, care must be taken not to cause pile-up
effects in strong, low-resolution reflections, whereby electrons
arriving on the same area in rapid succession will be undercounted
Low-Dose MicroED
315
all cause intensities to exhibit non-random variations over the
course of the measurement [35]. Ideally, all these effects are
accounted for during scaling, before intensities integrated from
the first observations are merged with those measured during
later stages. However, scaling a multicrystal dataset assumes that
all individual data originate from identical crystals, which is never
the case in practice. The effect of lattice discrepancies on the scaled
and merged intensities depends on the degree of non-isomorphism.
Multicrystal merging presents an opportunity to obtain a minimally damaged dataset, as frames recorded from a highly exposed
crystal can be discarded in favor of the first frames recorded from a
different sample. If enough datasets are available, merging a highmultiplicity dataset from short exposures can thus reduce errors and
lead to overall better results [36]. Even though MicroED often
requires data from a 10
–20
wedge for accurate indexing and
integration, not all frames need to be merged into the final dataset.
A good strategy for multicrystal data collection is often to alternate
between collecting data starting at high and low tilt. In cases of
preferred crystal orientation, this will increase the probability that
reciprocal space is evenly exposed, and that damage is not concentrated to specific wedges of reciprocal space. In the limit, only the
first frame from each dataset would be merged, which requires data
from very many crystals. Increasing automation now has the potential to make such a strategy feasible [37].
4 Camera Considerations
The quality of the recorded diffraction images is paramount in
MicroED, as they constitute the only experimental measurement
on which the final atomic model is based. Weak exposures generally
yield noisy images that cannot be accurately integrated, whereas
high exposures may fall outside the linear range of the camera. An
ideal camera for diffraction measurement would account for every
electron that interacts with the sample; it must have high dynamic
range in order to accurately record both the intense, low-resolution
reflections as well as the faint, high-resolution spots on the same
image.
The ability to accurately detect very low electron counts is
critical when the exposure is attenuated in the interest of reducing
radiation damage. A particularly attractive camera option is
provided by electron-counting, which owing to the absence of
read-out noise promises to accurately record weak, high-resolution
reflections. However, care must be taken not to cause pile-up
effects in strong, low-resolution reflections, whereby electrons
arriving on the same area in rapid succession will be undercounted
Low-Dose MicroED
315