We discuss the joint development by Penn State University (PSU) and Teledyne Imaging Systems (TIS) of hybrid CMOS detectors for X-ray astronomy, and specifically the development over the past 10 years of a new event-driven X-ray detector for future astronomy missions. This novel X-ray detector is designed to perform onchip event recognition and to read out only pixels containing X-ray events. With the exception of analog power supply voltages, the detector is digital in/digital out, reducing off-chip electronics to a minimum. It operates at frame rates of over 1000 frames per second, providing excellent performance for bright X-ray sources and/or high-throughput optics. The pixel size is 40 × 40 microns, and we are fabricating devices with 550 × 550 pixels.
In this paper, we present the test results of a flight-grade 13μm pixel pitch 6000-element 1.7μm InGaAs linear array in a hermetic package, designed and developed for space remote sensing and imaging applications. The array consists of a single 13μm pixel pitch 6000-element InGaAs linear array and a custom single digital 2.0 Mecapacitance trans-impedance amplifier (CTIA) readout integrated circuit (ROIC) with four gains. We have achieved greater than 80% peak quantum efficiency and higher than 1100 signal-to-noise ratio (SNR) at 90% well fill. The focal plane array is in a vacuum hermatically sealed package with an anti-reflective (AR)-coated Sapphire window and 29 pins, including four for low voltage differential signaling (LVDS) outputs.
A new burst-mode, 10-frame, hybrid Si-sensor/CMOS-ROIC FPA chip has been recently fabricated at Teledyne Imaging Sensors. The intended primary use of the sensor is in the multi-frame 800 MeV proton radiography at LANL. The basic part of the hybrid is a large (48×49 mm2) stitched CMOS chip of 1100×1100 pixel count, with a minimum shutter speed of 50 ns. The performance parameters of this chip are compared to the first generation 3-frame 0.5-Mpixel custom hybrid imager. The 3-frame cameras have been in continuous use for many years, in a variety of static and dynamic experiments at LANSCE. The cameras can operate with a per-frame adjustable integration time of ~ 120ns-to- 1s, and inter-frame time of 250ns to 2s. Given the 80 ms total readout time, the original and the new imagers can be externally synchronized to 0.1-to-5 Hz, 50-ns wide proton beam pulses, and record up to ~1000-frame radiographic movies typ. of 3-to-30 minute duration. The performance of the global electronic shutter is discussed and compared to that of a high-resolution commercial front-illuminated monolithic CMOS imager.
A high-resolution hybrid visible imager, that is composed of a CMOS readout integrated circuit (ROIC) and a silicon photo-detector array, has been designed. The ROIC is fabricated with a standard 0.25 μm CMOS mixed-mode process with a back-illuminated silicon detector array that is produced at Rockwell Scientific Company (RSC) using RSC's HyViSITM process.
The camera system is designed primarily to record images formed on a scintillator used in pulsed proton radiography experiments. In such experiments, the repetition rate of the proton beam can be as high as 2.8 MHz (358 ns). An imaging system with the desired 1440x1440 pixels resolution would result in an instantaneous readout rate in excess of 5.79 E12 samples/s. To address this issue we designed a pixel with three-frame in-pixel analog storage allowing for a deferred slower readout.
The 26 μm pitch pixel imager is operated in a global shutter mode and features in-pixel correlated double sampling (CDS) for each of the three acquired frames. The CDS operation is necessary to overcome the kTC noise of the integrating node to achieve high dynamic range. A 65 fps continuous readout mode is also provided. The hybridized silicon array has close to 100% fill factor while anti-reflection (AR) coating maximizes its quantum efficiency at the scintillator emission wavelength (~415 nm).
The ROIC is a 720x720, two-side buttable integrated circuit with on-chip 12-bit analog to digital converter (ADC) for digital readout. Timing and biasing are also generated on-chip, and special attention has been given to the power distribution of the pixel-array and snapshot signal buffers. This system-on-chip approach results in a compact and low power camera, an important feature to extend the number of imaged frames by synchronizing multiple cameras.
A 2D pixel array image sensor module has been designed for time resolved Protein Crystallography. This smart pixels detector significantly enhances time resolved Laue Protein crystallography by two or three orders of magnitude compared to existing sensors like films or phosphor screens coupled to CCDs. The resolution in time and dynamic range of this type of detector will allow to study the evolution of structural changes that occur within the protein as a function of time. This detector will also considerably accelerate data collection in static Laue or monochromatic crystallography and make better use of the intense beam delivered by synchrotron light sources. The event driven pixel array detectors, based on the column architecture, can provide multiparameter information (energy discrimination, time), with sparse and frameless readout without significant dead time. The prototype module consists of a 16 by 16 pixel diode array bump-bonded to the integrated circuit. Different detector materials (Silicon, CdZnTe) are evaluated. The detection area is 150 by 150 micrometers2 connected to the readout electronics. The individual pixel processor consists of a low-noise amplifier shaper followed by a differential threshold comparator which provides the counting of individual photons with an energy above a programmable threshold. To accommodate the very high rates, above 5 by 108/cm2/s, each pixel processor has a 3 bit pre-scaler which divides the event rate by 8. Overflow from the divider which defines a pseudo fourth bit will generate a readout sequence providing the pixel address. Addresses, generated locally as analog signals, are converted off-chip and used to increment a location in an histogramming memory to generate the computerized image of the Laue diagram.