smaller spread of intensities, due to the majority of the scattering atoms being close
in atomic number. The mosaicity of a protein crystal is generally higher, and the
diffraction limit is generally lower than that of a similarly sized crystal of a chemical
compound. These differences are key in understanding how detectors were developed for protein crystallography and are also informative in how chemical crystallography at synchrotrons developed in parallel. It could be considered that the
development of detectors suitable for this work presented the greatest initial challenge and it is only comparatively recently that detector technology is close to
catching up with the full potential that third-generation synchrotron sources have
to offer for crystallographic studies using single-crystal methods.
Dectris have developed a new series of pixel array photon counting detectors,
Eiger, which have significantly smaller pixel sizes (75 Â 75 μm rather than
172 Â 172 μm) and with a much faster 3 μs operating at a frame rate of up to
3 kHz. With this new breed of detector, it is now possible to perform serial
crystallography with wedges of data collected on several sample crystals per second
from a slurry of microcrystals spread on a sample plate. At a synchrotron with small
molecule crystals, there is the issue of encountering the limitations of the Dectris
HPC chip designs, in the Eiger and Pilatus detectors. Because of the thresholding
which is used to detect a single photon, a photon on the edge of two or more pixels
may not create enough charge in any one pixel to be detected. Because of the relative
sharpness of small molecule crystal peaks, this effect is more likely to affect small
molecule data. This issue is avoided in HPC detectors built on the MediPix chip
design [38], such as the X-Spectrum Lambda series, which allows adjacent pixels to
essentially compare and account for these below threshold events. Another issue
with all HPCs which is more commonly encountered in chemical and mineral
crystals is the pileup effect. Essentially only one photon can be detected at a time
in an HPC, and this creates a maximum photon rate that can be handled. Recently a
workaround has been implemented in the Pilatus 3 and Eiger detectors, but this has
only increased the limiting rate; it has not eliminated it. This pileup effect can lead to
intense low angle peaks being underestimated. Alternatively, Bruker has developed
the PHOTON series of detectors, based on CMOS technology [26] coupled to a
phosphor. This detector is a single monolithic charge integrating pixel array which
can simultaneously count and integrate incoming photons and therefore has no count
rate saturation point – it is limited by the pixel well depth and readout time.
Looking forward, synchrotron beamlines of all types will certainly be working
with brighter and more intense sources and will have to learn from the detectors of
the FELs [39, 40]. Photon counting is generally too slow for FELs – most FEL
detectors are integrating detectors. The construction is similar to HPCs, but the
readout chip architecture is designed to measure the charge on the sensor pixel,
not the arrival of individual photons. One of the current integrating detectors, like the
JUNGFRAU [41] from PSI at the SLS, has been shown to collect high-quality
protein crystallography data. The pileup issues seen in today’s detectors will not be
tolerable on future crystallography beamlines, when multiple orders of magnitude
more flux is arriving at the sample.
82
S. J. Coles et al.
in atomic number. The mosaicity of a protein crystal is generally higher, and the
diffraction limit is generally lower than that of a similarly sized crystal of a chemical
compound. These differences are key in understanding how detectors were developed for protein crystallography and are also informative in how chemical crystallography at synchrotrons developed in parallel. It could be considered that the
development of detectors suitable for this work presented the greatest initial challenge and it is only comparatively recently that detector technology is close to
catching up with the full potential that third-generation synchrotron sources have
to offer for crystallographic studies using single-crystal methods.
Dectris have developed a new series of pixel array photon counting detectors,
Eiger, which have significantly smaller pixel sizes (75 Â 75 μm rather than
172 Â 172 μm) and with a much faster 3 μs operating at a frame rate of up to
3 kHz. With this new breed of detector, it is now possible to perform serial
crystallography with wedges of data collected on several sample crystals per second
from a slurry of microcrystals spread on a sample plate. At a synchrotron with small
molecule crystals, there is the issue of encountering the limitations of the Dectris
HPC chip designs, in the Eiger and Pilatus detectors. Because of the thresholding
which is used to detect a single photon, a photon on the edge of two or more pixels
may not create enough charge in any one pixel to be detected. Because of the relative
sharpness of small molecule crystal peaks, this effect is more likely to affect small
molecule data. This issue is avoided in HPC detectors built on the MediPix chip
design [38], such as the X-Spectrum Lambda series, which allows adjacent pixels to
essentially compare and account for these below threshold events. Another issue
with all HPCs which is more commonly encountered in chemical and mineral
crystals is the pileup effect. Essentially only one photon can be detected at a time
in an HPC, and this creates a maximum photon rate that can be handled. Recently a
workaround has been implemented in the Pilatus 3 and Eiger detectors, but this has
only increased the limiting rate; it has not eliminated it. This pileup effect can lead to
intense low angle peaks being underestimated. Alternatively, Bruker has developed
the PHOTON series of detectors, based on CMOS technology [26] coupled to a
phosphor. This detector is a single monolithic charge integrating pixel array which
can simultaneously count and integrate incoming photons and therefore has no count
rate saturation point – it is limited by the pixel well depth and readout time.
Looking forward, synchrotron beamlines of all types will certainly be working
with brighter and more intense sources and will have to learn from the detectors of
the FELs [39, 40]. Photon counting is generally too slow for FELs – most FEL
detectors are integrating detectors. The construction is similar to HPCs, but the
readout chip architecture is designed to measure the charge on the sensor pixel,
not the arrival of individual photons. One of the current integrating detectors, like the
JUNGFRAU [41] from PSI at the SLS, has been shown to collect high-quality
protein crystallography data. The pileup issues seen in today’s detectors will not be
tolerable on future crystallography beamlines, when multiple orders of magnitude
more flux is arriving at the sample.
82
S. J. Coles et al.