New CCD detector for macromolecular diffraction data
Rick Walter
Section of Biochemistry, Molecular and Cell Biology, Cornell University
Higher intensity beamlines at synchrotron sources have driven the development
of new x-ray sensitive detectors with enhanced performance capabilities
for the collection of macromolecular diffraction data. Desired characteristics
include improved sensitivity, dynamic range, count rate, and efficiency
of data collection. To address this, a new integrating detector based upon
a silicon chip charge-coupled device, or CCD, has been developed and successfully
applied to the collection of macromolecular diffraction data at CHESS.
This detector is currently being made available to the macromolecular crystallographic
user community as part of the upgraded data collection capabilities of
the newly recommissioned A1 line. Dan Thiel and I have helped users collect
their first data on the detector. In its early applications, the detector
has proven exceptional for the collection of macromolecular diffraction
data.
The CCD detector has been developed as a collaboration between Don Bilderback
of CHESS, the laboratory of Dr. Steven Ealick at Cornell, the group of
Dr. Sol Gruner of Princeton University (including Sandor L. Barna, Michael
E. Wall) and Dr. Eric Eikenberry of the Robert Wood Johnson Medical Center
and Dr. John Lowrance of Princeton Scientific Instruments. Both a schematic
and a photograph of the detector are shown.
Photograph of the CCD detector in place at the CHESS A1 station. A
protective metal screen (shown open at left) protects the phosphor when
not collecting data .
Operationally, x-ray photons are converted to visible photons by a
Gd2O2S:Tb phosphor. The pixel size at the phosphor end is 50 mm x50mm.
This signal is transferred from the phosphor by a 2.6:1 fiber optic taper
to a 1024 x1024 pixel CCD chip (~20 mm on an edge). Signal is stored in
the chip by the generation of about 11 electrons per 13.6 keV x-ray absorbed,
which is then maintained by an imposed voltage until readout. At readout,
charge stored in a pixel row is sequentially shifted horizontally to the
single readout amplifier. Subsequently, the charge in each row of pixels
above is shifted downward and read out in like manner. In this way, a stationary
controller converts the horizontal and vertical position of each pixel
and its stored charge to a digitized x,y coordinate on the detector face
and a raw intensity. These data are then written to some sort of storage
device (currently a magnetic disk or tape) and displayed on a monitor for
inspection and limited manipulation.
In practice, this detector has provided several advantages for the collection
of macromolecular diffraction data. To begin, the CCD chip pixel size is
19.46mm which translates roughly into 50 mm square pixels at the detector
face. With a bloomed point spread function at full width half maximum of
80 mm, the detector truly gives a spatial resolution limited by the pixel
size. Further, detector sensitivity at wavelengths from 8 keV to 14 keV,
as measured by fraction of x-rays stopped, ranges from 99% to 70% efficiency,
respectively. Finally, the readout dynamic range of the detector is 36,000
x-rays. In its current configuration, a 20 second CCD readout time is used.
While this is comparatively slow for a chip of this size, it was chosen
as a compromise between minimizing readout noise from the single amplifier
and decreasing the duty cycle for each macromolecular diffraction image.
(This time will be reduced in future detectors.) In addition to such performance
capabilities, each image is immediately available for inspection and limited
manipulation which includes image zoom and spot intensity integrating and
profiling capabilities, among others. While the active area of the detector
is rather small, its spatial resolution allows crystal to detector distances
as close as 35 mm to be used (giving 1.4е data on an edge with 0.91е wavelength
x-rays in a symmetric configuration). For medium to slightly larger unit
cell dimensions (high resolution data for cells with axes up to 180е have
been successfully collected), the detector can be offset and pulled back
to give more reasonable overlap while still recording high resolution data
(<2.0е)
Schematic of the CCD detector (Sol Gruner).
Relative to crystallographic results, an example of a diffraction image
from data collected using this detector on A1 is shown at below. During
a three day experimental run in September, seventeen complete data sets
were collected for five different protein crystallography projects. The
diffraction image is from a frozen crystal of a complex between the enzyme
nucleoside deoxyribosyl transferase and a substrate, deoxyuridine (cubic,
I213 crystals with unit cell axes of 148.2е). The Rsym for all data was
5.0% to 2.3е, and it is currently being used for high resolution refinement
and active site analysis. Because of detector sensitivity and speed and
flash freezing of the crystals (see page 44), a complete (97%), high quality,
high resolution data set was possible from a single crystal. This was previously
impossible with these extremely radiation sensitive crystals.
Diffraction image from a crystal of a complex between the enzyme nucleoside
deoxyribosyl transferase and deoxyuridine, a substrate. Data was usable
to 2.3е in this data set. The image is a 20 second, 0.75° oscillation with
a crystal to detector distance of 60 mm (this detector configuration allowed
2.2е data on the edges and 1.7е data in the corners of the detector).
A second example produced an electron density difference for the active
site of the enzyme bovine purine nucleoside phosphorylase in complex with
a non-cleavable substrate analogue, 9-deazainosine (cubic, P213 crystals
with unit cell axes of 93.3е). Again, high quality, high resolution data
with an Rsym of 5.2% for all data to 1.7е (94% complete) were obtained
using the CCD detector in conjunction with flash freezing. These data are
currently being used to investigate the mechanism of the enzyme. Such results
were typical of the data collected during this run. Since then, the detector
has been used to collect thousands of images during a month and a half
long MAD phasing run on the F2 beamline and is currently being applied
to the collection of macromolecular data for outside users on A1.
At CHESS, the development of beamlines and improved x-ray detectors, such
as the CCD, is going hand in hand. Plans for continued improvement in detector
development are already in place. These include doubling the active area
and decreasing the readout time to five seconds in the current design and
a planned mosaic detector design. The CCD detector and planned improvements
to it will play a role in keeping CHESS at the forefront of macromolecular
crystallography for the foreseeable future.