X-ray Imaging Spectrometer (XIS)
Suzaku at ISAS
XRS at GSFC
HXD at Tokyo U.
XRT at GSFC
The X-ray Imaging Spectrometer (XIS) is one of three science instruments flown on the Suzaku X-ray satellite. There are four independent XIS sensors aboard the spacecraft, each with its own telescope and electronics, and each employing an X-ray sensitive CCD. These sensors are the latest in a line of development that includes the ASCA SIS and Chandra ACIS detectors. Each XIS CCD contains 1024 by 1024 pixels and provides a square field of view 17.8' on a side. Each pixel subtends about 1", and the angular resolution of the XIS is determined by the ~2' half-power diameter of the X-ray Telescopes (XRTs). Energy resolution ranges from 50-180 eV (FWHM) over the 0.2-10 keV energy range. One of the sensors is a thinned backside-illuminated (BI) device, allowing substantially improved sensitivity at energies below 2 kev. The XIS devices are quite similar to the CCDs flown on Chandra ACIS, but various improvements increase the spectral resolution of the BI chip and mitigate the effects of on-orbit radiation damage.
This page focuses on some of the technical innovations made for the XIS detectors. For complete practical information for proposers, see the XIS Chapter of the Suzaku Technical Description and the Suzaku Technical Documents posted by ISAS.
General Comparison to Chandra ACIS
The general characteristics of the XIS are compared to those of ACIS in the table below. The first part compares the CCD sensors themselves, while the second part compares the parameters determined by typical observing conditions (i.e., all 4 XRT+XIS paths for Suzaku, ACIS-I or ACIS-S configuration for Chandra). All values are pre-flight with the exception of the the ACIS effective areas, which include the effects of OBF molecular contamination as of 2004.
Charge Injection for Better Radiation Tolerance
The primary instrument aboard Suzaku, the XRS, has a 30-36 month life before its cryogen is exhausted. Because of this, focus for the first part of the mission will be driven by XRS-specific science, and the XIS will need to perform well late into the mission, after the XRS is no longer functional. Some radiation damage from high-energy particles is inevitable given the satellite's orbit (about 600 km altitude). This damage creates CCD lattice traps that capture charge during readout, reducing the charge transfer efficiency (or increasing the charge transfer inefficiency, CTI). CTI results in a gain shift, as charge from all photon events is lost and the resultant PHA and inferred energy are lower. In addition, the charge loss per transfer depends on the signal charge (photon energy) and location on the chip (rows near the readout encounter fewer traps than the top rows); when combined with the stochastic nature of the charge trapping, this leads to broadening of the response function at a given energy and thus a degradation of the spectral resolution.
To mitigate the effects of radiation damage, the XIS CCDs were developed so that signal can be injected into the top row of the chip. This charge injection signal is tunable to any energy within the usable passband. Charge injection mitigates radiation damage in two ways: first, it reduces the effects of CTI by filling the traps with "sacrificial charge" during readout; and second, it allows for on-orbit measurement of CTI with much less overhead than similar calibration done with X-ray sources.
To alleviate CTI during observations, charge can be injected periodically into the top row of the array during readout (see Figure 1). When the chip is next read out, the injected charge fills the traps, preventing subsequent rows from losing signal. If the trap release time constant is greater than the injection period, then the traps remain full, signal charge from X-ray events is not lost, and CTI is reduced. Tests on a flight spare CCD irradiated with 40 MeV protons show that an injection period of about 100 rows (2.4 ms) is sufficient to keep the traps occupied with sacrificial charge. Such a scheme reduces the usable area of the CCD by 3% (including the effects of event grading), however it increases the spectral resolution by about 30% and markedly reduces the gain shift compared to using a damaged chip without charge injection (see Figure 2). In particular, the spectral resolution with charge injection is within 10% of the pre-irradiated value.
Charge injection also enables tracking of CTI changes with time, energy and location. Such calibration measurements can be done with X-ray sources, as they have for Chandra/ACIS, however this requires considerable observing time to obtain the millions of photon events required, well-distributed in energy. Charge injection can be tuned to produce levels from 50-3000 electrons/pixel, equivalent to about 0.3-15 keV/pixel, with precision better than the noise of an X-ray event of similar amplitude. Each energy can be measured in the time it takes to read out the chip. In practice, multiple (10-12) rows are injected in sequence. The amplitude of the first (bottom) row after readout measures the CTI, while the amplitude of the final (top) row indicates the reference energy level, as intermediate rows have filled the majority of traps with sacrificial charge. Column-to-column CTI variations are easily measured in this way. Since charge can only be injected at the top of the array, row-to-row variations are not recoverable. At the very least, charge injection provides a very overhead-cheap complement to X-ray CTI measurements.
At this point in the Suzaku mission, the protocol for using charge injection has yet to be determined.
Improvement: Back Illuminated CCD Process
Back-illuminated CCDs have a higher QE than front-illuminated (FI) detectors with the same depletion depth. This is especially true at low energies where the very thin deadlayer on the illuminated surface absorbs few photons compared to the thicker deadlayer and gate structure of the FI devices. Until recently, however, poor charge collection near the surface has resulted in reduced energy resolution compared to FI CCDs, which are near the theoretical limit.
A new fabrication technique has been developed to prevent charge loss in thinned back-illuminated CCDs. This chemisorption charging process, developed by Dr. Michael Lesser and colleagues at the University of Arizona (Lesser & Iyer 1998), results in thin layers of HfO and SiO2 on the illuminated surface, and increases the spectral resolution at low energies to nearly the theoretical limit without compromising the QE. This is demonstrated in Figure 3 which shows a comparison of the QE and spectral resolution for the BI and FI chips.
For this reason, the single BI XIS represents an improvement over previous generations of BI devices. Figure 4 shows simulated spectra of the supernova remnant E0102-72.3, using a model based on data obtained by the Chandra/HETG instrument. The top figure shows the count rate expected from lines of helium- and hydrogen-like oxygen and neon using a single BI and FI XIS, using ground-based values for the telescope transmission parameters. The BI produces up to three times the FI count rate, at similar resolution. Simulations of the Chandra/ACIS-S BI devices and XMM-Newton/EPIC-PN are shown in the middle and bottom plots, respectively. The spectral resolution of the BI XIS is significantly better than the other instruments.
Note: Some of these papers make reference to two BI chips being flown on Suzaku. The actual flight complement was later changed to 3 FIs and 1 BI.
XIS Chapter of the Suzaku Technical Description