Page 187 - Macromolecular Crystallography

176 MACROMOLECULAR CRYS TALLOGRAPHY

typical crystalline protein samples, thereby wast- using smaller sample oscillations/rotations at the
ing most of the photons emerging from the undu- expense of recording more diffraction images over
lator X-ray source. Therefore, the X-ray beam is a greater period of time. We have found that this
focused to match the size of the crystal. On undu- approach may not always eliminate the problem of
lator beamlines, focusing is usually achieved either overlaps. In addition, such fine slicing introduces
through sagittal focusing, using specially designed a non-trivial complication to subsequent data pro-
monochromator crystals, or with the aid of a pair cessing. Use of smaller oscillation/rotation ranges
of bendable mirrors. In a typical Kirkpatrick–Baez reduces the number of reflections that are fully
focusing system (Kirkpatrick and Baez, 1948) two recorded on a single diffraction image, which makes
mirrors are used to compress the beam, one in the it more difficult to scale together data from different
vertical dimension and the other in the horizontal. images. At SGX-CAT, it has proven more effective
At SGX-CAT, such mirrors yield an X-ray beam that to maintain as large an oscillation range as pos-
is approximately 70 µm × 80 µm at the sample. sible and increase the sample-to-detector distance.
This approach increases the distance between reflec-
tions on the face of the CCD detector. The sizes
of the reflections remains essentially unchanged
12.3.2 Detectors
because of the collimated X-ray beam produced by
Chapter 5 described the types of detectors available the undulator insertion device.
for in-house data collection. In-house diffraction A new generation of pixel-array detectors is cur-
experiments generally rely on image-plate or CCD rently under development for use in protein crys-
detectors, whereas virtually all modern protein crys- tallography (Brönnimann et al., 2004, 2006). These
tallography synchrotron beamlines use only CCD detectors exhibit minimal point-spread factors. In
detectors (Fig. 12.2). Although image-plates provide addition, their extremely rapid readout, <10 msec,
a greater dynamic range, the time required for data permits diffraction data to be acquired on a near
read out is too long relative to the synchrotron expo- continuous basis.
sure time. For example, readout for a Rigaku/MSC
RAXIS IV ++ is 50 to 100 sec, with a minimum total 12.3.3 Crystal handling
cycle time of 3 min, including reset of the image
plate (Fujisawa et al., 2003). The fastest CCD detec- Acquisition of the diffraction images is only one part
tors have a readout time of less than 2 sec, which is of the data collection process. Cryogenically pre-
comparable to typical exposure times used at third- served crystals have to be placed in the X-ray beam,
generation synchrotrons (0.4 to 12 sec). The speed of while being maintained at liquid nitrogen temper-
these detectors, combined with the short exposures atures. There has been a considerable effort in the
available at the synchrotron, results in very rapid development of robotic systems to transfer crystals
data collection. With current technology, a dataset from a storage dewar to the goniostat for data col-
containing 180 diffraction images can be recorded lection. Custom systems have been developed at
easily in less than 9 min. various synchrotrons for this purpose (Cohen et al.,
CCD detectors use a phosphor to convert the 2002; Snell et al., 2004; Jacquamet et al., 2004; Pohl
incoming X-ray to visible light, which is in turn et al., 2004; Karain et al., 2002; Ueno et al., 2004). This
detected by the CCD chip. During this conversion variety of robots reflects the fact that each beamline
process, the apparent size of an X-ray reflection tends to be of unique design. As automated sam-
increases, a phenomenon known as the point-spread ple changing becomes more established, it is likely
factor. Typically, this change in reflection size does that beamline robotics will reflect a greater degree
not present a significant challenge. In extreme cases of standardization. Three commercial systems are
(i.e. large unit cell dimensions or a very short sam- now available from Rigaku (Abad-Zapatero, 2005)
ple to detector distance), however, it can lead to (based on an earlier design from Abbott Labo-
overlaps between adjacent reflections. When over- ratories (Muchmore et al., 2000)), Bruker, and
laps do occur, diffraction data can be recorded MAR Research.