Overview
One of the often neglected necessities of synchrotron radiation science is detector technology. The Gruner Detector Group designs and builds novel detectors for various applications including macromolecular x-ray crystallography, microcrystallography, x-ray tomography, and electron microscopy. Our design goals include high-speed imaging, high spatial resolution, wide dynamic range, compatibilty with femtosecond signal pulses, and high application versatility for storage ring, energy recovery linac, and x-ray free electron laser synchrotron sources, as well as for non-synchrotron sources. See below for a list of current projects.

Pixel Array Detectors

Pixel Array Detectors (PADs) are being developed by the Detector Group to meet the x-ray detection needs of the synchrotron community. In overview, a PAD is an application-specific integrated circuit (ASIC) coupled to a diode detector via bump-bonding. The ASIC offers a flexible platform for development, as well as a greater potential for high speed imaging via commercial grade mixed-mode CMOS circuits. In addition, the diode itself directly converts absorbed x-rays into electron hole pairs, which are swept into the ASIC for processing by an applied electric field. Compared to common scintillator-based imaging arrays that must detect secondary photons produced by the scintillation material, a PAD's direct detection method offers much better signal-to-noise ratios and spatial resolution

To the right is a basic illustration of a PAD chip (not to scale). The top layer (blue) is the diode used for converting absorbed x-rays to charge. The diode is typically silicon, but other materials with shorter absorption lengths (e.g. GaAs) can be used for more efficient detection of harder x-rays. The bottom layer (green) is the CMOS ASIC. This is an application-specific integrated circuit custom-made to the requirements of the experiment. The diode and the ASIC are mated using bump-bonds (represented by the balls between the the two). On the right hand side of the detector the I/O to the ASIC is seen. The dimensions of the detector pixels depend heavily on the application; typical values are between 100 and 200 microns on a side.

CCD detector technology that resulted from earlier work by the group is in use at every hard x-ray synchrotron source in the world. PADs resulting from work by the group have been or are being used at synchrotron x-ray sources around the world.
Basic illustration of a PAD chip Basic illustration of a PAD chip


Projects

Pixel Array Detectors
LCLS CSPAD
The Cornell-SLAC pixel array detector (CSPAD) is the first PAD designed for a hard X-ray Free Electron Laser (XFEL), developed for use at at SLAC (Linac Coherent Light Source (LCLS)). Variants of the CSPAD now serve as the primary detectro instruments at the LCLS.

An XFEL produces femtosecond pulses of coherent x-rays with fluxes that are orders of magnitude above that produced with other man-made x-ray sources. Since the x-rays from each pulse arrive in femtoseconds, photon counting PADs were not useful, and a Cornell integrating PAD design was chosen. The CSPAD achieved a signal-to-noise ratio of 8 and a dynamic range of >2500 for 8 keV x-rays. The detector was designed as a 3-side buttable chip of 194x185 pixels, each 110 microns on a side.

The Gruner group designed and fabricated the CSPAD chips, while SLAC built them into large detectors. Three full scale detectors (1516x1516 pixels) and about a half dozen 758x758 pixel detectors are now operational at the LCLS.

References: 225, 243, 246, 267, 274, 282, 28f; SLAC publication

Mixed Mode PAD (MMPAD)
This detector is designed to have a very wide dynamic range without losing single x-ray sensitivity. This is done by designing a pixel that has combined analog and digital character, hence the name mixed-mode PAD. The scheme is to use an architecture with noise that may grow, but is always much less than the equivalent shot noise in the x-ray signal.

The current MMPAD detector is a 3-side buttable detector of 128x128 pixels, each 150 microns on a side. It has a SNR of about 6 for 8 keV x-rays, frames at 1 kHz, and has a dynamic range of >3x107 x-rays per pixel per frame. A 2x3 module detector is already in use at Cornell, and more are being built. Currently the chip is being mechanically redesigned to allow assembly of arbitrarily many chips into a mosaic with chip gaps of only a few millimeters. The MMPAD is also being adapted for use in electron microscopy.

An even wider dynamic range version, the Hihg Dynamic Range PAD (HDR-PAD) is being developed.

References: 196, 226, 265, 281, 289, 294, 301, 302, 308, 317, 319, 25f, 28f

Keck PAD
Designed for microsecond time-resolved X-ray imaging, this PAD is the successor to the well-known 100x92 prototype PAD that has been used for fuel injector and reactive metal foil research. It was designed for single bunch imaging at the APS with successive full-frame (14-bits) images every 150 ns, and even faster for fewer bits.

The Keck PAD
is designed to capture a frame every ~150 ns. It uses an integrating architecture to store the signal of eight successive frames on an on-chip, per-pixel, array of holding capacitors. A 16x16 pixel chip has already been tested; construction of a 2x3 array of 128x128 pixel chips is underway.

References: 205, 250, 264,303,318,322, 28f

FPGA PAD
This PAD is being designed to move functionality from hardware into firmware. The idea is to mate a FPGA to each PAD ASIC (application-specific integrated circuit) via a wide-bandwidth pathway. The functionality of the ASIC depends on the firmware in the FPGA. Thus, a single ASIC can accomplish different PAD functions.

References: 284

Electron Microscope PAD (EM-PAD)
This is a variant of the MM-PAD adapted for use as the detector in scanning transmission electron microscopy (STEM). It allow rapid data collection of the full 4-dimentional diffraction patterns of objects in STEM mode. It also enables all the standard STEM microscopies.

References: 242, 320, 323, 328, 329

General Detector Considerations and PADs for Hard (> 15 keV) X-ray Imaging

References: 216, 258, 287, 290, 306, 311, 314, 315
sCMOS X-ray detector
The sCMOS detector was developed for high spatial resolution, high energy applications. It is a small (5mm x 5mm) detector capable of expanding into a large format, with <10 micron resolution and good detection efficiency for x-rays in the 25-100keV range. It uses scintillating fiber optics to convert x-rays to length so as to have high x-ray stopping power while preserving spatial resolution. A fiber-optic magnifying taper couples the initial optics to a sCMOS chip.

References: 253, 291, 31f

Direct Detection CCD
This detector is a large area (10 cm x 10 cm), large format (4k x 4k pixels) direct detection charge-coupled device (CCD) for macromolecular crystallography. This is a relatively inexpensive alternative to PADs that will outperform phosphor-coupled CCDs in low flux applications, such as microcrystallography.

The detector consists of a CCD structure on the front-side of a thick Si high-resistivity wafer. The wafer is back-biased to fully deplete it so the entire thickness of the wafer is sensitive to x-rays; this provides good x-ray stopping power up to 20 keV. Each stopped x-ray produces thousands of electron-hole pairs. While this quickly fills the pixel well-depth, limited to about 106 electrons, thereby limiting x-rays stored to each pixel to a few hundred, the signal per x-ray should greatly exceed the per pixel read noise, so the detector quantum efficiency should be excellent. To compensate for the limited well-depth, the CCD is divided into 16 ports that all read out in parallel at high speed.

The final device frames at 2-5 Hz with noise low enough to see single 12 keV x-rays. The direct-detection CCD performs competitively with the more common phosphor-coupled fiberoptic taper CCD detectors.

References: 293, 32f


People


PADs in the Press

Shock Wave Movies


Presentations and Selected Publications



Last updated 2016-10-26 SMG