Making focal plane detectors a reality for next-generation hard x-ray telescopes will require an imaging spectrometer that counts high-energy photons from the invisible part of the electromagnetic spectrum (10–50keV). In the field of astronomy, charge-coupled devices (CCDs) with exquisite energy resolution (down to ∼140eV at 6keV) have been widely used for x-ray imaging below 10keV. Efforts have been made to extend the energy coverage by developing so-called fully depleted CCDs with >300μm thickness. Yet these devices still would not provide the needed efficiency to detect photons in the desired energy band. Moreover, as a detector becomes thicker, background noise caused by cosmic-ray particles from space increases. Adding an active anticoincidence shield (another kind of detector) helps to mitigate this problem. Because cosmic-ray particles leave signals in both the image sensor and the shield, the timing coincidence of the signals between the two detectors can be used to identify and remove the noise. But this approach requires better timing resolution (<1μs) than is possible with CCDs to efficiently exploit the anticoincidence.
Figure 1. Cross-sectional view of a DSSD. The highly doped positively charged or p-type silicon strips (p+: yellow) and the negatively charged or n-type silicon strips (n+: black) are implanted orthogonally to provide two-dimensional coordinate measurements. Each n+ strip is surrounded by a floating p+-doped implantation to be isolated from any adjacent strips. Aluminum (Al) electrodes are directly coupled on each strip with ohmic contact.
An alternative candidate for the new hard x-ray detectors, with more than three orders of magnitude greater timing resolution, is the double-sided silicon strip detector (DSSD). This device was created to track charged particles in high-energy physics. The basic geometry of the strip, which includes orthogonally implanted n and p strips on both sides of the detector (see Figure 1), provides two-dimensional coordinate measurements for x-ray photons and other absorbed particles. The 500μm-thick device is practical and has ∼40% photoabsorption efficiency for 20keV. Stacking a set of fully depleted DSSDs improves performance. For example, the efficiency obtained with 20 layers (for a total thickness of 1cm) is 70% or higher for 40keV imaging.
An additional issue for the DSSD at this frequency is noise performance. The signal generated by a hard x-ray photon is much smaller than that made by a charged particle. For this reason, we wished to minimize electrical noise in the DSSD and readout electronics. Because strip capacitance is the greatest contributor to this problem, we took the preferred approach to reducing the noise, which is to widen the strip gap and narrow the strip width. But a wider gap also increases the risk of p-n junction breakdown owing to a higher electrical field concentration at the end of the junction. The strip pitch for one of these devices within a telescope can vary from 100 to 400μm, depending on the plate scale and the point spread function of the optical instrument.
We set an upper limit of 100μm on the strip gap to lessen the possibility of junction breakdown. Similarly, optimizing the DSSD thickness involves a tradeoff between interaction efficiency and depletion voltage. Therefore, we decided on a DSSD thickness of 500μm to keep the operation voltage below 200V.
Another key technology in keeping superfluous data to a minimum and achieving the required energy resolution is low-noise front-end electronics. We developed the ASIC (application-specific integrated circuit) VATA series in collaboration with Gamma Medica–Ideas, Norway.1 The geometry of the front-end field effect transistor and the shaping time (∼4μs) has been optimized for input capacitance of 6–10pF and leakage current of less than 0.1nA, with a power constraint of 0.2mW/channel. The typical noise performance is 50e− (RMS) at 0pF load and 180e− at 10pF, which corresponds to 1.4keV full width at half maximum (FWHM) in silicon with 2μs shaping time. The electronics channel has been DC−coupled to the silicon strip to avoid degradation of the noise performance due to extra components associated with the AC coupling.
Figure 2 shows a 133Ba line image in the energy band from 20 to 40keV. The classic−style car mask, made of 0.3mm thick brass, was mounted 3mm above the DSSD. The detector used here was 4cm wide, with 300μm thickness and 400μm strip pitch manufactured by Hamamatsu Photonics, Japan.2 For testing purposes, the thickness of the this DSSD was thinner than the optimum. The position resolution had been confirmed to be consistent with the strip pitch of the DSSD. Figure 3 shows the sum of the 241Am energy spectrum for all 96 strips on p strips. The energy resolution was 1.5keV (FWMH), which is consistent with the value calculated from the VA32TA noise performance and input capacitance (∼12pF).
Figure 2. A shadow image using x-rays from 133Ba (barium) in an energy band from 20 to 40keV.
Figure 3. Spectrum performance of the DSSD irradiated with 241Am (americium).
We next devised a method to stack the detectors. Some form of stacking is essential to obtain enough efficiency in the higher-energy band. Figure 4 shows a prototype of a DSSD stack module. It has four newly developed DSSD boards, which consist of one 2.56cm-wide and 300μm-thick DSSD as well as a readout ASIC mounted on each side of the detector. These boards are stacked with a 2mm pitch. All layers operated successfully with a high-energy resolution of 1.6keV (FWHM).3 We also checked the improved efficiency of the stack, which corresponds to a total thickness of 1.2mm (300μm×4 layers). Note that this technology is scalable, so that a total 1cm thickness is practicable with 10–20 layers.
Figure 4. DSSD stack module consisting of four layers with a pitch of 2mm.
Overall, we have demonstrated successful performance of a low-noise DSSD system. The resolution characteristics of this system are attractive for high-quality hard x-ray imaging spectroscopy. We also successfully fabricated a high-density compact DSSD stack, which is critical for greater efficiency. The flight model development of the DSSD system for the hard x-ray imager onboard the NeXT (New exploration X-Ray-Telescope) mission,4 the planned sixth in a series of Japanese x-ray astronomy satellites, is a continuing process. Imaging detectors in the 10–100keV range would be welcome in fields such as astrophysics, medicine, and nondestructive inspection. To this end, development of a next-generation Compton gamma-ray telescope is now under way.5,6
Shin'ichiro Takeda, Tadayuki Takahashi, Shin Watanabe
Institute of Space and Astronautical Science/Japan Aerospace Exploration Agency (ISAS/JAXA)
Shin'ichiro Takeda is a PhD student and a Japan Society for the Promotion of Science (JSPS) research fellow at ISAS/JAXA and the University of Tokyo.
Tadayuki Takahashi is professor of physics at ISAS/JAXA and the University of Tokyo. He is leading an international team that is developing NeXT, planned to be launched in 2013.
Shin Watanabe is assistant professor of physics at ISAS/JAXA. He has been involved in research related to silicon and cadmium telluride semiconductor detectors for x-ray and gamma-ray astrophysics.
Hiroyasu Tajima, Takaaki Tanaka
Stanford Linear Accelerator Center
Hiroyasu Tajima is a staff scientist. He is a member of the GLAST (Gamma-ray Large Area Space Telescope) satellite and principal investigator of the soft gamma-ray detector onboard the NeXT satellite.
Takaaki Tanaka is a JSPS postdoctoral fellow working in high-energy astrophysics.
University of Tokyo
Kazuhiro Nakazawa is a lecturer in the Department of Physics working in high-energy astrophysics. He is one of the core team members of the hard x-ray detector onboard the Suzaku x-ray satellite.
Yasushi Fukazawa is associate professor of physics at Hiroshima University. He has been involved in a series of Japanese x-ray astronomy satellites as a core member of the instrument team.