Overview and Current Status of the ASCA SIS
--by G. R. Ricker, MIT
The ExperimentThe Solid State Imaging Spectrometer (SIS) on ASCA consists of two camera assemblies (SIS-0 and SIS-1) and dedicated analog electronics units (AE). Digital data handling is provided by the spacecraft digital electronics/digital processor unit (DE/DP). Compared with the GIS, the SIS has improved energy resolution (4 times better at most energies), greater low energy (Ex < 1.5 keV) area, and better angular resolution, while the GIS excels in effective area at high energies (Ex > 5 keV), field of view (4 times greater solid angle), and high countrate capability. The SIS hardware was jointly developed by the Massachusetts Institute of Technology (MIT), ISAS, and Osaka University.
The two SIS cameras are identical, with each enclosing four 420 x 422 MIT Lincoln Laboratory CCD chips, edge-abutted to comprise an 840 x 848 focal plane. The CCDs are frontside illuminated. Individual pixels are 27 µm x 27 µm. The focal plane coverage of each SIS camera extends over 22 arcmin x 22 arcmin. In flight, the CCDs are cooled to a stabilized operating temperature of -60¡ C. The energy resolution at 5.9 keV is 2%, or ~120 keV FWHM; at 1 keV, it is since their depletion regions extend to a depth of ~30 µm.
The SIS sensors function by collecting the charge from direct photoelectric interaction of X-rays with the silicon atoms in the CCDs themselves; no indirect energy converters or charge multiplication by high voltages are involved. Background rejection in the SIS relies upon both charge amplitude discrimination (i.e. a minimum ionizing particle produces charge equivalent to a 15 keV X-ray, beyond the critical reflection energy for the ASCA XRT mirrors) and charge topology discrimination (i.e. particle interactions are characterized by tracks extending over several pixels, or by "blobs" of diffuse charge collected from low field regions of the CCD). The selection of valid SIS X-ray interacions is performed partly onboard ("event" selection, "grade" assignment) and partly on the ground (pixel screening,grade summation). Details of the SIS event screening algorithms are given in the ASCA Research Announcement (NRA 93-OSS-08), and in a forthcoming paper (Ricker, et al 1994). Since the SIS cameras represent the very first in-orbit use of single photon-detecting CCDs, their success is one of the major technical accomplishments of the ASCA mission. As was hoped prior to launch, the SIS CCDs consistently demonstrate excellent energy resolution and good quantum efficiency combined with extremely low non-X-ray background (< 5x10-4 cts cm-2 s-1 keV-1) over a broad energy range (0.4 - 10 keV).
Following the launch of ASCA on February 20, 1993, the SIS cameras were permitted to outgas for several weeks, primarily through a delayed-action, self-activated vacuum vent. During the period from 17-29 March 1993, the two cameras were fully activated electrically, and their entrance doors were opened to accept the full range of X-rays reflected from the ASCA XRTs. From 30 March to 6 April, initial in-orbit testing and preliminary calibration took place. On 15 April, performance verification (PV) observations began. Over the PV period (15 April to 15 October 1993), the SIS cameras continued to perform stably and successfully. In general, supernova remnants (e.g. Puppis A, Cas A, W49B, and Vela X) were utilized for line and low energy calibrations, while cluster sources (e.g. the Coma Cluster) and a QSO (3C273) have been used as continuum references.
Overall PerformanceThe SIS cameras on the ASCA satellite continue to perform extremely well.The SIS instrumental parameters (temperatures, bias voltages, clock voltages, etc.) were quite stable though out the ASCA PV phase (e.g. CCD temperatures were stable to ± 0.05 C). As a result, features as fine as 5-10 eV are discernible in a number of high signal-to-noise spectra! The demonstrated limiting sensitivity (5 sigma level of confidence) achieved by the SIS in a 105 sec observation is ~1.0 x 10-3 counts/sec in a fixed 3 arcminute diameter detection cell in the 0.4-10 keV bandpass. The principal issues affecting camera performance are discussed in the sections below.
Gain and Readout NoiseThe SIS CCD gains have remained near ~1 e-/ADU, just as they were prior to launch. Similarly, the SIS system readout noise has remained within the 4-6 e- RMS range measured prior to launch. The following table gives the specific values on a chip-by-chip basis, measured in flight using orbit nighttime, "corner pixel", 4 CCD Faint Mode data:
Readout Noise (e- RMS) Linear Gain (e- per ADU) Camera Chip0 Chip1 Chip2 Chip3 Chip0 Chip1 Chip2 Chip3 S0 5.30 5.19 5.41 6.02 0.9768 0.9792 0.9010 0.9503 S1 4.03 4.12 6.29 4.94 0.8733 0.8958 0.9638 0.9399
Hot and Flickering Pixels (HFPs)An area of concern for the SIS is the continued increase with time of the telemetred rate of hot and flickering pixels (HFPs). (The number of such pixels is not large per se -- less than 1% of each CCD is so effected. However, when the HFPs are active, they can fill the available telemetry in 4 CCD, Low Bit Rate mode, pre-empting bandwidth, which otherwise could be used for transmitting valid events.) In 4-CCD Faint mode, the rate had risen to 16 and 19 cts s-1 for Sensor S0 and S1, respectively, as of October 1993. Several corrective measures currently are under consideration by the SIS team. Through active measures, it may be possible to repair the radiation damage responsible for the spurious events. (Laboratory measurements at MIT are planned to test a number of such measures on ASCA SIS flight spare sensors.) Alternatively, it should be possible to raise the event threshold of the detector passively, trading off low energy spectral coverage for telemetry bandwidth.
Optical Contamination and Light LeaksOperationally, a major concern has been to minimize the degree of optical contamination of the CCD data by light leakage from the sunlit earth during orbit day. This contamination effect was well-studied during the PV phase; orbital conditions that tend to produce the light leakage have been identified and fed back into the SPIKE observation planning. In addition, an area discriminator now is applied routinely on the two chips most affected for observations where this otherwise would be an issue.
SIS BackgroundThe internally generated, non X-ray background (NXB) for the SIS has been estimated using data obtained when ASCA is viewing the earth on the nighttime portion of the orbit (so-called "dark earth" data). For regions of high magnetic rigidity (12-50 Gev/c, occurring over ~50% of the ASCA orbit), the NXB rate for the SIS cameras is keV -1, averaged over the 0.4-10 keV energy band. For regions of lower rigidity (0-12 Gev/c) the NXB rate increases to ~7 x 10-4ÊctsÊcm-2Ês-1ÊkeV-1. These values are ~2x lower than previously reported for the SIS, and have been achieved through careful attention to eliminating hot and flickering pixels from each chip during ground-based analysis. In all, ~1500 pixels/chip or ~1% of those in each array, are rejected in the analysis. By comparison, the diffuse x-ray background (DXB) contributes ~3.5x10-3 cts cm-2 s-1 keV-1Êaveraged over the 0.4Ñ10 keV band, with the bulk of the DXB counts occurring in the 0.4Ñ3 keV range.
Quantum efficiencyThe measured quantum efficiency (QE) of the two SIS cameras combined with the XRT, is ~16% at 0.5 keV, near the lower limit (~0.4 keV) of their energy range. The SIS + XRT QE peaks at ~84%, near 3.5 keV; at 6 keV, the QE of the combined SIS + XRT is ~50%. For comparison to the GIS and to other missions, the implied count rate for the Crab Nebula, assuming a photon index = 2.05 and a column density of nH = 3x1021 atoms cm-2, is 770 cts s-1 for a single SIS camera. [NB: The Crab Nebula itself is actually too bright for the SIS to observe directly without excessive event pileup. As yet, we have not established a single, stable photometric astronomical X-ray standard of appropriate brightness usable by the SIS over its entire energy range. Instead, we have been forced to "synthesize" standards using various astronomical targets in combination with pre-launch, ground-based calibration data.]
Changes in the low energy sensitivity of the SIS have been monitored at periodic intervals following launch. In principle, the low energy sensitivity could be altered by the presence of contaminants that accumulate on the cold SIS focal planes as the rest of the ASCA satellite outgasses. No significant variations have been measured between periodic observations of Vela-X, a low extinction, continuum source. The sensitivity of this study has been limited by counting statistics, and a 2% change in sensitivity over the 0.30-0.55 keV band within four months would correspond to a measurement significant at the 1 sigma level. However, no significant change in the low energy sensitivity has been measured. The upper limit for the contamination rate set for SIS Sensor 0, Chip 1, corresponds to an accumulation rate (1 sigma) of <1.2x10-6 g cm-2 yr-1 [=6x1016 atoms(C) cm-2 yr-1]. As the ASCA mission progresses, we should be able to refine our measurement of this contamination rate even further. However, no anomalous low energy absorption in SIS source spectra have been seen, and spectral fits to previously well-studied objects yield consistent neutral hydrogen column densities within expected errors. Thus, as far as we can tell, the low energy sensitivity of the SIS detectors is stable.
Stability of the SIS Camera GainsThe SIS-DE/DP system generates a pulse height for each photon-induced SIS event. Prior to post-telemetry correction for pixel "echoes" and dark-frame errors, pulse heights may include systematic errors of up to 1% at 2 keV. Following correction analysis, the systematic errors are reduced to less than 0.2% over the entire energy range. The systematic gains have been stable within 0.2%. PV-phase, orbit nighttime observations generally have yielded hydrogenic line positions within this tolerance. Orbit daytime data is subject to additional uncertainties, which we expect to resolve soon. We have been looking closely for changes in the gain, possibly due to charge transfer inefficiency induced by radiation damage. From the first six months of PV phase data, only upper limits of 3 x 10-5 (2-sigma level of significance) can be placed on the charge transfer inefficiency (CTI) of the CCD arrays. This value is near the limit of predictions for cumulative radiation damage expected in the ASCA orbit (Gendreauet al 1993), so that measurements based on 9-12 months of ASCA SIS data may begin to show the effects of CTI.
Energy Calibration and Response MatricesThe energy calibration for one chip on each sensor (SIS0-Chip1 and SIS1-Chip3) now has been fully checked in flight; the remaining in-flight chip calibrations and checks are still in progress. The energy scale has been found to have slight nonlinearities near the Si edge and at high energies. The SIS detector response at high energies has been constrained by the GIS observations of 3C273, which find it to be a featureless power law with a photon index of 1.60. Using existing data, we thus have been able to fix the response of two of the chips; the high energy response of the remaining six chips awaits a calibration observation scheduled for completion later in December. The low energy response of all chips has been tentatively established using pre-launch ground calibration data. The current generation of matrices are available through the use of the (ver.2.7) FTOOL SISRMG or through the ISAS response tool SIS_BLDRSP.
Current calibration activities are oriented primarily toward completion of the calibration program outlined above. In addition, a number of observations of bright (high extinction) sources shows a small (~1%) redistribution of high energy events into a low energy pulse height tail. We are investigating a number of candidate mechanisms that might be held responsible for the effect. Another area to which we shall be devoting considerable effort is defining the appropriate corrections to the particle-induced errors in the energy scale. Typically, the dark-frame update mechanism that is responsible for fixing on-orbit the zero point of the pulse height scale is downshifted by one or two electrons, depending on observing conditions. The effect can be corrected fully for faint mode data. For brighter sources, this effect must be handled by the response matrices. Finally, we expect to be making improvements in the dead layer models that are used to describe the low energy detection efficiency. This action should enable us to remove some of the current uncertainty in the chip-to-chip variations of detection efficiency, which are expected to be on the order of ~5-10 %.
Return to the Table of Contents Proceed to the next article