next up previous contents
Next: Bibliography Up: Suzaku Technical Description Previous: 6. X-Ray Telescopes (XRT)   Contents


7. X-Ray Imaging Spectrometer (XIS)

7.1 XIS Overview

7.1.1 XIS Basics

The X-ray Imaging Spectrometer (XIS, Koyama et al., 2007) celebrated its first light on 2006, August 11, about a month after the launch of Suzaku. The instrument has been operated successfully since then, producing many scientific results. The entire instrument has a weight of 48.7kg and consumes 67W at a bus voltage of 50V during normal operations. The XIS is composed of four units of Si-based X-ray charge coupled device (CCD) cameras (Fig. 7.1). In these X-ray sensors, incident X-ray photons are converted into a number of electron-hole pairs via photoelectric absorption and subsequent ionization by photoelectrons and their secondaries. The energy of the incident photon can be measured since it is proportional to the amount of charge produced.

Each XIS unit is located at the focal plane of one of four independent, identical, and co-aligned X-Ray Telescope modules (XRT, Serlemitsos et al., 2007), which are called XRT-I0, XRT-I1, XRT-I2, and XRT-I3. The XIS is operated in photon-counting mode, in which each X-ray event is discriminated from the others and its position, energy, and arrival time are reconstructed. This gives the XIS imaging-spectroscopic capabilities in the 0.2-12.0keV energy band over an $18^{\prime}\times18^{\prime}$ region.

Figure 7.1: A photo of one of the four XIS sensors before installation on the satellite.
\includegraphics[totalheight=3in]{figures_xis/xis} Similar instruments

The XIS is similar to its predecessor, the Solid-state Imaging Spectrometer (SIS) onboard the ASCA satellite, in its working principle. Many improvements were made based on successful use of the SIS over eight years. It is also similar to its brothers: Chandra ACIS (Garmire et al., 2003), XMM-Newton EPIC (Turner et al., 2001; Strüder et al., 2001), and Swift XRT (Burrows et al., 2004).

X-ray CCD instruments, including the XIS, are characterized by their flexibility in operation and possible rapid performance changes on orbit. In particular, micro-meteorite hits can leave unrecoverable damage to a part of the instrument. The entire imaging area of the XIS2 was lost in 2005 November and a part of the XIS0 in 2009 June. Merits

The XIS has several advantages over other instruments. Responsible parties

The XIS was developed and has been maintained jointly by Japan and the United States, with participating organizations including the MIT, ISAS, Kyoto University, Osaka University, Rikkyo University, Ehime University, Miyazaki University, Kogakuin University, Nagoya University, Aoyama Gakuin University, and major participating contractors including the Mitsubishi Heavy Industries (MHI), the NEC-Toshiba Space Systems, and the System Engineering Consultation (SEC) Co. Ltd.

7.1.2 XIS Components Sensors Dimensions

Each CCD camera has a single CCD chip, which consists of the imaging area and the frame store area. The imaging area is exposed to the sky for observations, while the frame store area is shielded. Each chip is composed of four segments called segment A, B, C, and D. Each segment has its own readout node (Fig. 7.3).

Figure 7.2: Field of view of the XIS. All sensors are co-aligned.

The imaging area has a pixel size of 1024$\times $1024 pixels. The pixel scale is 24$\mu $m pixel$^{-1}$ and the physical size is 25mm squared. The plate scale is 1.04 $^{\prime\prime}$ pixel$^{-1}$ and the total sky coverage is 18$^{\prime}$ squared.

Figure 7.3: Schematic view of the data flow in one XIS unit.
\includegraphics[height=6.0in]{figures_xis/xis_config_v2} CCD types

The XIS1 is a back-side illuminated (BI) chip. The other three sensors (XIS0, 2, and 3) are front-side illuminated (FI) chips. The BI and FI chips are superior to each other in the soft and hard band responses, respectively. Optical blocking filter (OBF)

X-ray CCDs are also sensitive to optical and UV photons. To suppress such signals, an optical blocking filter (OBF) is installed on the surface of the CCDs. The OBF is made of polyimide with a thickness of 1000Å, which is coated with a total of 1200Å of Al (400Å on one side and 800Å on the other). Calibration sources

For in-flight calibration, three $^{55}$Fe calibration sources with a half life of 2.73 years are installed in each XIS unit. The $^{55}$Fe sources emit strong Mn K$\alpha $ and K$\beta$ lines at 5.9keV and 6.5keV, respectively. Two sources illuminate corners of the segments A and D of the imaging area at the far side of the readout node. The other one is installed in the door. It was used to illuminate the entire chip before opening the door for the final check before launch and the initial check after launch. Because the door was opened for observations, the door calibration sources are not used any longer. Scattered X-rays from the door calibration sources may appear in the entire imaging area. Cooling System

To reduce the dark current, the sensors are kept at $\sim-90^\circ$C all the time. Thermo-electric coolers (TECs) using the Peltier effect are used for cooling, which are controlled by the TEC Control Electronics (TCE). Electronics Analogue electronics

The analog electronics (AE) system drives the CCD by providing driving clocks for exposure and charge transfer, sampling the voltage, amplifying the data, and converting to the digital values. The XIS has two identical units of the AE and the TCE (see, Cooling system control), which are stored in the same housing for each unit. One unit (AE/TCE01) is used for XIS0 and XIS1, while the other (AE/TCE23) is used for XIS2 and XIS3. Digital electronics

The digital electronics (DE) system processes the digitized data. The DE has two pixel processing units (PPU01 and PPU23) and a main processing unit (MPU). The digital data from the AE are stored in the Pixel RAM of the PPUs; PPU01 for data taken with AE/TEC01 and PPU23 for AE/TEC23. The PPUs access the raw CCD data in the Pixel RAM, carry out event detection, and send event data to the MPU. The MPU edits the telemetry packets and sends them to the satellite's main digital processor.

7.1.3 CCD Pixels & Coordinates RAW XY coordinates

Pixel data collected in each segment are read out from the corresponding readout node and sent to the Pixel RAM. In the Pixel RAM, pixels are given RAW X and RAW Y coordinates for each segment in the order of the readout, such that RAW X values are from 0 to 255 and RAW Y values are from 0 to 1023. These pixels in the Pixel RAM are named active pixels. In the same segment, pixels closer to the read-out node are read out earlier and stored in the Pixel RAM earlier. Hence, the order of the pixel read-out is the same for segments A and C, and for segments B and D, but different between these two segment pairs, because of the different locations of the readout nodes. In Fig. 7.3, numbers 1, 2, 3 and 4 marked on each segment and the Pixel RAM indicate the order of the pixel read-out and the storage in the Pixel RAM. In addition to the active pixels, the Pixel RAM stores copied pixels, dummy pixels and H-over-clocked pixels (Fig. 7.3). At the borders between two segments, two columns of pixels are copied from each segment to the other. Thus these are named copied pixels. Two columns of empty dummy pixels are attached to the segments A and D. In addition, 16 columns of H-over-clocked pixels are attached to each segment. ACT XY coordinates

Actual pixel locations on the chip are calculated from the RAW XY coordinates and the segment ID during ground processing. The coordinates describing the actual pixel location on the chip are named ACT X and ACT Y coordinates (Fig. 7.3). It is important to note that the RAW XY to ACT XY conversion depends on the on-board data processing mode.

7.1.4 Major Events

The detector performance changes both continuously and discontinuously over time. This is especially the case for CCD instruments. Users need to take the latest information into account in order to make the best use of the instrument. Some major changes and their causes are:

Table 7.1 shows major events since the start of the mission in chronological order. The XIS log at gives a complete list of events relevant for data reduction.

% latex2html id marker 2964\begin{table}[hp]
...ntly for saving power.\\
\end{longtable} \end{table}\end{landscape}

7.2 XIS Observation Planning

7.2.1 User Options Available options

The XIS has numerous instrumental settings, which are fine-tuned to serve a wide range of observational purposes. A few of them are left as user options. Different options can be used for different sensors.

  1. Clocking mode
  2. Editing mode
  3. PPU ratio
  4. Lower event threshold
  5. Area discrimination

In many cases, the default settings are adequate. If this is not the case, the appropriate choice of clocking mode options is generally enough and there is little need to change the other options. The choice between the XIS and HXD nominal positions is no longer available: all observations are made at the XIS nominal position (the center of the XIS field of view) unless otherwise requested. Observations requiring non-default settings

Users need to consider an appropriate choice of clocking mode options when they are planning observations of bright ($>$10mCrab in 0.5-10keV) sources or those requiring a higher temporal resolution than 8s. Declaration of the use of non-default setting

Use of non-default options should be stated clearly in the technical justification of the proposal document. The choice of clocking mode needs to be specified in the cover page. These choices can be tentative. For successful proposals, an inquiry is sent to the PI about three weeks prior to the observation. The choice of user options can be revised at this time. Last-minute changes

The behavior of very bright sources is often unpredictable. The XIS operation team waits until the last minute before finalizing the options, if necessary. The deadline is 10:00 JST (1:00 in UT) every weekday and Saturday, at which the operation team starts generating the command sequence. Users need to submit the final choice of options by this deadline 1-3 days before the start of their observation. Prior consultation with the XIS operation team is necessary for their availability and the exact deadline. Consultation

The appropriate choice of options is the responsibility of the observer. A summary is provided by the XIS quick reference at:$\sim$tsujimot/pg_xis.pdf
The XIS operation team can be consulted at:
The team will make the best use of the operational flexibility to maximize the scientific output of the observations. Clocking Modes

The clocking mode specifies how the CCD pixels are read out. Each clocking mode has its own $\mu $-code, a pattern of voltage clocking for exposure and charge transfer. It enables full read, partial read, or stacked read (Table 7.2). With a partial or stacked read, a higher pile-up limit and timing resolution can be achieved at a sacrifice of observing efficiency and imaging information.

Table 7.2: Clocking modes

\begin{tabular}{\vert lllccc\vert}
...the window options.\par
\end{footnotesize} \end{tablenotes} \end{threeparttable}

There are two major types of clocking mode: the Normal mode and the P-sum mode. The Normal mode is for timed-exposure readout. It can be combined with a window option (partial readout in space), burst option (partial readout in time), or both (partial readout in both space and time). Fig. 7.4 shows the time sequence of exposure, frame-store transfer, readout, and storage to the Pixel RAM (in the PPU) for the Normal clocking mode. The P-sum mode is for stacked readout. The P-sum clocking mode is in principle available for all FI sensors. However, because it is severly affected by leaked charge in areas damaged by micrometerorite hits, the P-sum clocking support was terminated for the XIS2 and XIS0. Therefore, it is currently available only for the XIS3. Observers using the P-sum clocking mode for the XIS3 need to use a Normal clocking mode for the other sensors.

Figure 7.4: Time sequence of the exposure, transfer, and readout. For the window option, 1/4 window is assumed.
\includegraphics[height=5in]{figures_xis/xis_option} Normal clocking mode

  1. No option: full readout of all pixels with a frame time of 8s.
  2. Window option: partial readout in space. In the 1/$w$ window mode, the central 1024/$w$ pixels are read out in the Y direction and the entire pixels are read out in the X direction, yielding a 1024$\times $1024/$w$ pixel image. The exposure time is 8/$w$ s.
  3. Burst option: partial readout in time. In the $b$s burst mode, all pixels are read out, but the effective exposure time is limited to $b$s ($b<$8s). During 8s of exposure time, events detected in the first ($8-b$)s are transfered and discarded without being recorded. Events detected in the remaining $b$s are transfered and recoreded.
  4. Window$+$burst option: partial readout in both time and space. In 1/$w$ window and $b$s burst mode, the central 1024$\times $1024/$w$ pixel image is read out with an exposure time of 8/$w$s. During the exposure time, events in the first ($8/w-b$)s are discarded and those in the last $b$s are recoreded. The inequity of 8/$w>b$ always holds.

The available window and burst options are summarized in Table 7.3.

Table 7.3: Window and burst options (XIS nominal position)

\begin{tabular}{\vert l\vert c\vert cc\vert cccc\vert cc...
...ransfer of 156 ms.\par
\end{footnotesize} \end{tablenotes} \end{threeparttable} Some advice in selecting window and burst options P-sum clocking mode

The pulse height from 128 rows are stacked into a single row along the Y direction. Charge is transferred continuously, thus the spatial information along the Y direction is lost and replaced with the arrival time information. The arrival time is ticked at a 7.8ms resolution. This is smeared by the point spread function with a HPD of 2$^{\prime}$, which is equivalent to 0.9s. Editing Modes

The editing mode specifies the telemetry format of the events. The XIS has several editing modes (Table 7.4). Four of them are for observational purposes, the remaining ones are for diagnostic purposes.

Table 7.4: Editing modes and size per event for observation modes

\begin{tabular}{\vert lllccc\vert}
Purpose & Cl...
...nding on the grade.\par
\end{footnotesize} \end{tablenotes} \end{threeparttable} Editing modes for observations

For Normal clocking modes three editing modes (5$\times $5, 3$\times $3, and 2$\times $2) are available. The 2$\times $2 mode is used only for the FI sensors. For P-sum clocking mode, the editing mode is fixed to the timing mode. Each editing mode has a different telemetry format, which includes the x and y position and the pulse height (PH) of the event (Fig. 7.5). Different editing modes have different sizes per event (Table 7.4).If the dark-subtracted PH value is above the split threshold in the neighboring pixels, they are considered to be a part of the charges generated by the event and thus are summed in the event reconstruction. Different thresholds are applied for the immediately neighboring pixels (inner split threshold) and those around them (outer split threshold).

Figure 7.5: Data pattern for 5$\times $5, 3$\times $3, and 2$\times $2 editing modes. 1-bit information indicates whether the pulse height of the pixel exceeds the outer split threshold.

  1. 5$\times $5 mode: The PH values of each 5$\times $5$=$25 pixel around the event center are sent to the telemetry.
  2. 3$\times $3 mode: The PH values of each 3$\times $3$=$9 pixel around the event center and the 1-bit information (whether the PH is larger than the outer split threshold or not) for the surrounding 16 pixels are sent to the telemetry.
  3. 2$\times $2 mode: The PH values of each 2$\times $2$=$4 pixel around the event center and the 1-bit information (whether the PH is larger than the outer split threshold or not) for the attached 8 pixels are sent to the telemetry. The 2$\times $2 pixels are selected such that they include the pixels with the highest, the second highest, the third highest (,and the fourth highest) PH values.
  4. Timing mode: The summed PH value of the 1$\times $3 pixels around the event center and the event grade are sent to the telemetry. The PH is summed if the neighboring pixels have a PH value above the inner split threshold.

Although the amount of information for event reconstruction is different among the 5$\times $5, 3$\times $3, and 2$\times $2 modes, no significant difference has been found so far between the former two modes. Only a small difference is seen for the 2$\times $2 mode for its lack of ability to perform trail correction (§ The 2$\times $2 mode and the timing mode are not available for the XIS1 because the trailing correction does not work properly for the former mode and the amount of flickering pixel is larger than for the other sensors for the latter mode. Editing mode combinations

Technically speaking, arbitrary combinations of editing modes for the four sensors are possible. However, for reasons of operational reliability, combinations are restricted to those listed in Table 7.5. Editing modes for diagnastics

The diagnositic modes are only for diagnostic purposes and not available to general users. The XIS is operated in these modes outside of observing times.

  1. Frame mode: The pulse height data of all pixels are dumped.
  2. Dark frame mode: The dark levels of all pixels are dumped. A dark frame is taken once every day during an SAA passage. PPU Ratio

The PPU ratio controls the relative fractional telemetry allocation for the four sensors. The ratio is fixed to be the optimum value assuming the use of the Normal clocking mode with the same options for all the sensors (5x5, 3x3, or f2x2_b3x3 combination) or the use of the P-sum clocking mode for XIS3 and the Normal clocking mode with the 1/4win $+$ 0.5s burst option for XIS0 and 1 (s3tim_s015x5, s3tim_s013x3, or s3tim_s02x2_s13x3 combination).

Table 7.5: Combination of editing modes and PPU ratio

\begin{tabular}{\vert lllllll\vert}
Mode & 5x5 ...
...he PPU ratio value.\par
\end{footnotesize} \end{tablenotes} \end{threeparttable} Lower Event Threshold

The lower event threshold defines the pulse height above which an event is recognized. It practically controls the lowest bandpass of the detector. The default event threshold is shown in Table 7.6. For XIS calibration observations, the FI event threshold is set to 50.

Table 7.6: Event threshold
Sensor Event threshold
  (ADU) (keV)
FI 100 0.365
BI 20 0.073 Area Discrimination

A part of the image can be masked with an area discrminator. Either the inside or the outside of a single rectangular region can be applied as a mask separately for each sensor. This is useful to supress telemetry of unwanted bright sources within the field of view of the target source. Currently, an area discrimination mask is applied permanently to the XIS0 to suppress leaked charge from the area damaged by a micrometeorite hit.

7.2.2 Systematic Observational Effects and their Mitigation Photon Pile-Up Description

Photon pile-up is caused when multiple photons arrive at a detection cell within the frame time. If two photons with energies of $E_{1}$ and $E_{2}$ arrive, they are wrongly recognized as a single photon with an energy of $E_{1}+E_{2}$. This causes underestimation of the count rate, spectral hardening, PSF distortion, grade migration, and various other effects. Photon pile-up is a concern in observing point-like sources brighter than $\sim$10mCrab. The degree of pile-up is measured by the pile-up fraction. Assume the mean count rate of photons landing in a pixel is $\lambda$s$^{-1}$pixel$^{-1}$. Then, the mean number of counts ($\overline{n}$) in a detection cell ( $A_{\mathrm{cell}}$) within the frame time ( $t_{\mathrm{frame}}$) is

\overline{n} = \lambda A_{\mathrm{cell}} \Delta t_{\mathrm{frane}}.
\end{displaymath} (7.1)

The number ($n$) fluctuates around $\overline{n}$ following Poisson statistics. The probability to have $n=k$ photons in a detection cell within the frame time is

P(n=k; \overline{n}) = \frac{\overline{n}^{k} e^{-\overline{n}}}{k!}.
\end{displaymath} (7.2)

The pile-up fraction (PF) is defined as the fraction of pile-up events among all incoming events as

= \frac{\sum_{k=2}^{\infty} P(n=...
= 1-\frac{\overline{n}}{e^{\overline{n}}-1} .
\end{displaymath} (7.3)

Obviously, $\mathrm{PF}(\overline{n}) \rightarrow 0$ as $\overline{n}
\rightarrow 0$ and $\mathrm{PF}(\overline{n}) \rightarrow 1$ as $\overline{n} \rightarrow \infty$.

$\lambda$ is a function of the position in the PSF $f_{\mathrm{PSF}}(x,y)$ as

\lambda(x,y) = N_{\mathrm{total}} f_{\mathrm{PSF}}(x,y) \Delta x \Delta y
\end{displaymath} (7.4)

where $N_{\mathrm{total}}$ s$^{-1}$ is the total count rate of a source and $\Delta x \Delta y$ is the area of a pixel. $\lambda(x,y)$ is largest at the center $\lambda(x=0,y=0)$ and becomes smaller as the distance from the PSF center ( $r=\sqrt{x^{2}+y^{2}}$) increases in the outskirts. Limits

The pile-up limit is defined as the total count rate ( $N_{\mathrm{pileup}}$s$^{-1}$) at which the PF($\overline{n}$) at the center of the PSF exceeds 3%. In the XIS, this happens at roughly 12s$^{-1}$, or 10mCrab. The limit should be regarded as an approximate value as the exact value depends on the spectral shape of the observed target. The tolerance also depends on scientific goals. For example, if the detection of a weak hard tail is the goal, the pile-up effect should be minimized. If the detection of a line is the goal, some degree of pile-up can be tolerated. Mitigation

Figure 7.6: Incident versus observed count rates for a point-like source. The ratio of the two gives the observing efficiency. The thick lines indicate the range below the pile-up limit. Telemetry saturation is applicable only to Normal clocking modes.

Table 7.7: Total count rate (s$^{-1}$) to cause the indicated pile-up fraction as a function of PSF radius.
Radius Pile-up fraction
  3% 20% 50%
$\sim$5 13.4 30.0 45.9
5$\sim$10 17.6 43.3 67.7
10$\sim$15 25.6 64.3 101
15$\sim$20 33.8 85.4 134
20$\sim$25 42.6 108 171
25$\sim$30 55.8 142 224
30$\sim$35 70.6 180 285
35$\sim$40 88.2 226 356
40$\sim$45 100 256 404
45$\sim$50 119 304 480
50$\sim$55 141 362 571

There are two major strategies to mitigate the effects of pile-up. They can be used together. One is to raise the pile-up limit by using an option in the Normal clocking mode. The pile-up limit for various options is summarized in Table 7.3 and shown in Fig. 7.6. The other is to only use events from the outskirts of the PSF, where the incoming rate per pixel is substantially smaller than in the center. This is called the annulus extraction method and is discussed in detail in Yamada et al. (2011). The tools are available at$\sim$yamada/soft/XISPileupDoc_20120221/XIS_PileupDoc_20120220.html.
Table 7.7 gives the total incoming rate, at which some representative pile-up limit is hit at various radii in the PSF. In this approach, the fraction of grade 1 events is known to be a good indicator to assess the degree of pile-up. Note that the attitude fluctuation of the satellite does not mitigate the photon pile-up as its typical time scale is much longer than the frame time of 8s. Telemetry Saturation Description

The XIS has a quota for the size of telemetry per unit time. The quota depends on the data rates (DR; SH$=$super high, HI$=$high, MED$=$medium, and LOW$=$low) and whether the observation is conducted on weekends or weekdays (Table 7.8). The satellite is also operated with a weekend allocation for a few days at the end and the beginning of the year. The operations team is responsible for deciding on the data rates depending on the observing conditions. The schedule is made to avoid bright sources to be observed on weekends, during which the telemetry allocation is small. Possible excess use of the quota is determined every 8s during observations. Beyond the limit, a part of the telemetry is lost. Events in the segments B and C are prioritized over those in the segments A and D. Telemetry saturation occurs only when observing bright sources. Fortunately, in the XIS, the telemetry saturation is always avoided when users select an appropriate clocking mode to avoid pile-up. In practice, the following users need to consider the possibility of telemetry saturation seriously:

Table 7.8: Telemetry allocation for XIS

\begin{tabular}{\vert llcc\vert}
DR & Condition...
...-off rigidity, etc.\par
\end{footnotesize} \end{tablenotes} \end{threeparttable} Limits

The telemetry saturation limit depends on the DR, choice of clocking modes, editing modes, and the PPU ratio. The resultant telemetry saturation limits are shown in Table 7.9 for the six editing mode combinations. A 10% margin for the headers and the background rates of 10s$^{-1}$ (FI) and 20s$^{-1}$ (BI) are included. The exact limit depends on the choice of clocking mode options. Users can use the XIS limit calculator at$\sim$tsujimot/limits_xis.ods.

Table: Incoming rate to cause telemetry saturation (s$^{-1}$)a

\begin{tabular}{\vert lrrrrrr\vert}
Edit mode &...
\end{footnotesize} \end{tablenotes} \end{threeparttable} Mitigation

The operations team is fully responsible for an appropriate choice of editing mode combinations among those in Table 7.5. The PPU ratio is optimized for a wide range of cases. Therefore, not much can be done by users in the observation planning stage to avoid telemetry saturation. In the data analysis step, however, users can remove time intervals afflicted by saturation by filtering the events using the distributed ``non-saturated'' good time interval files. The detailed procedure can be found in section 6.5.1 of the Suzaku Data Reduction Guide (ABC Guide). Out-Of-Time Events Description

Out-of-time events are events recorded during charge transfer. They spread along the Y direction beyond the PSF with low surface brightness. For accurate photometry, users need to take out-of-time events into account. It is expected that they have the same spectrum as the on-source events. The effect is present for the Normal clocking mode with any option, but is different for those with and without burst option. For the Normal clocking mode without burst option, the out-of-time events spread uniformly along the Y direction at a surface brightness of 156ms/(8-0.156)s$=$2.0% integrated over 1024pixels. For the Normal clocking mode with a $b$s burst option, the out-of-time events spread uniformly in the upper and lower half of the image with different strengths (Fig. 7.7). This is because the clocking of the $b$s burst options consists of several steps: (1) exposure for $\sim$(8-$b$)s, (2) readout and flushing charges, (3) exposure for $b$s, (4) readout and recording. The flushing readout is performed without charge injection and takes 25ms. The recording readout is with charge injection and takes 156ms. The different strengths in the upper and lower half of the image are caused by this. The fraction of out-of-time events is $\sim$25ms/$b$/2s in the upper part and $\sim$156ms/$b$/2s in the lower part. The fraction can be significant for short burst options, i.e., for the 0.1s burst option, the out-of-time events amount to 12.5% in the upper half and 78% in the lower half.

Figure 7.7: XIS1 image showing out-of-time events in an obersvation of the Crab at the HXD nominal position taken with a 0.1s burst option.
\includegraphics[totalheight=3in]{figures_xis/xis_oot_img} Mitigation

The contribution of out-of-time events can be estimated in the area far from the center of the image. This is difficult for observations taken with a window mode. In such cases, at least one of the sensors can be operated without a window option so that it can be used for estimating the out-of-time events also for other sensors. Self Charge Filling Description

In principle, charge traps can be filled not only by artificially injected charges, but also by charges created by X-ray events. This effect is apparent in observations of some very bright sources (Todoroki et al., 2012). This effect is not included in the calibration. Users may encounter a better energy resolution than the distributed RMF files indicate.

% latex2html id marker 3308\begin{table}[p]
...d{footnotesize} \end{tablenotes} \end{threeparttable} \end{table}\end{landscape}

7.3 XIS Performance & Calibration

7.3.1 Spaced-Row Charge Injection (SCI)

X-ray CCD devices are subject to degradation in orbit. One of the outcomes is an increase of charge traps under the constant radiation of cosmic rays in the space environment. This results in an increase of the charge transfer inefficiency (CTI), which leads to a degradation of the energy resolution.

Figure 7.8: Frame dump image of XIS2 with SCI. The bright lines every 54 rows are rows with injected charges. Other events are mostly due to cosmic rays. The gaps between the segments are due to over-clock sampling.

The XIS has a function to precisely monitor and mitigate this effect. For monitoring each sensor has $^{55}$Fe calibration sources at two corners on the far-side of the readout. For mitigation spaced-row charge injection (SCI) is implemented. Electrons are injected artificially from one side of the chip and are read out along with charges produced by X-ray events. The artificial charges are injected periodically in space (every 54 rows; Fig. 7.8). They fill up charge traps and thereby alleviate the increase in CTI for charges by X-ray events (Ozawa et al., 2009; Bautz et al., 2004; Uchiyama et al., 2009; Nakajima et al., 2008). SCI is not available for the P-sum clocking mode for its continuous clocking nature. CTI Monitoring & Trend

Fig. 7.9 shows the long-term trend of the measured peak energy and width of the Mn I K$\alpha $ line (5.9keV) from the calibration sources. The peak decreases and the width increases gradually. Both quantities are restored by applying SCI.

Figure 7.9: Trend of $^{55}$Fe peak height and width. These are raw values without trail correction. See Fig. 7.11 for the gain and energy resolution for calibrated data.
\includegraphics[width=1.0\textwidth]{figures_xis/cal_src} SCI Operations History Start of SCI

The SCI technique was put into routine operation in the middle of 2006 and has brought a drastic improvement. At the start of SCI operations, it was decided to inject the amount of charges equal to the amount produced by a 6keV X-ray photon (``6 keV equivalent'') for the FI devices and a smaller amount (``2 keV equivalent'') for the BI device. The smaller amount for XIS1 was chosen to minimize the expected SCI-related increase in noise in the soft spectral band, at which the BI device has an advantage over the FI device. Choice of SCI on/off

The choice between SCI on and SCI off was a user option for nearly one year after the start of SCI operations. This was because SCI observations suffer from a larger dead area and a larger fraction of out-of-time events in addition to providing an improved spectroscopic performance. The user option was terminated and all observations have been made exclusively with SCI on since AO4, as the number of users choosing SCI off had decreased substantially. SCI Increase for XIS1

The accumulation of contaminating material on the surface of the CCDs made the soft-band advantage of the BI sensor less prominent (§ 7.3.3). The CTI for the BI device has increased at a faster rate due to the smaller amount of injection charges. As a consequence, the astrophysically important lines of Fe XXV (6.7keV) and Fe XXVI (7.0keV) became hardly resolved in 2010. The injection charge amount was changed from 2keV to 6keV equivalent for the BI sensor according to the schedule of Table 7.1. Detailed information can be found in the Suzaku memo 2010-07 available at SCI Notes for Users & Proposers

Users need to be aware of some drawbacks associated with the SCI operations:

7.3.2 Energy Gain and Resolution

The calibration of the spectroscopic performance of the XIS mainly consists of calibrating the energy gain and resolution. Regarding the energy gain, corrections for charge trails and for the CTI are performed.

For the energy resolution, the time-dependence is modeled as

\Delta E(\mathrm{PH_{0}}, t) = \sqrt{a(t)+b(t)\mathrm{PH_{0}}+c(t)\mathrm{PH_{0}}^2},
\end{displaymath} (7.5)

where $\mathrm{PH_{0}}$ is the pulse height of the incident X-rays and $t$ is the time from the launch. The time variable parameters $a(t)$, $b(t)$ and $c(t)$ are determined phenomenologically using the $^{55}$Fe calibration sources and the E0102$-$72 observations.

For the normal clocking modes, data taken with the 3$\times $3 and 5$\times $5 editing modes are merged, since it is known that their differences are negligible. Energy Calibration Results (1) -- Normal Mode (No Options)

The Normal mode without options is the best calibrated mode of the XIS. The CTI is measured using observations of the Perseus Cluster (all segments, hard band), 1E0102$-$72 (segments B and C, soft band), and the $^{55}$Fe calibration sources (segments A and D, hard band). Fig. 7.10 and 7.11, respectively, show the gain and energy resolution in the hard band using the $^{55}$Fe calibration sources, while Fig. 7.12 and 7.13, respectively, show the gain and energy resolution in the soft band using the E0102$-$72 observations.

Figure 7.10: Gain in the hard band for the Normal mode (without options) using the $^{55}$Fe calibration sources. The upper panel shows the data before turning on the SCI and the lower panel after. The CTI is corrected.

Figure 7.11: Energy resolution in the hard band for the Normal mode (without options) using the $^{55}$Fe calibration sources. The upper panel shows the data before turning on the SCI and the lower panel after. The CTI is corrected.

Figure 7.12: Gain in the soft band for the Normal mode (without options) using E0102$-$72 observations. The upper panel shows the data before turning on the SCI and the lower panel shows after. The CTI is corrected.

Figure 7.13: Energy resolution in the soft band for the Normal mode (without options) using the E0102$-$72 observations. The upper panel shows the data before turning on the SCI and the lower panel after. The CTI is corrected.
\includegraphics[height=0.95\textheight]{figures_xis/fig16_2_OLyaWidth} Energy Calibration Results (2) -- Other Clocking Modes Normal mode (window option)

Since the $^{55}$Fe calibration sources are unavailable for observations with window options, the Perseus Cluster is regularly observed with the 1/4 window option for calibration purposes (Table 7.10). The target is observed in Normal mode twice, without option and with 1/4 window option, with consecutive exposures, in order to construct a comparison data set at each epoch. No calibration sources are observed with the 1/8 window option. The pulse height correction for the Normal mode with window option is the same as that for the Normal mode without options, except that they use different numbers of fast and slow charge transfers. Fast and slow transfers are used for reading out the effective imaging area and the non-imaging area, respectively. They have different CTIs. For the window options, the non-imaging area includes the frame store area as well as the imaging area outside of the window. The CTI for each transfer is calibrated using the calibration observations with the 1/4 window option, which is also applied to the 1/8 window option. Fig. 7.14, 7.15, and 7.16 show the resulting gain using several emission lines, comparing the Normal mode without options and the Normal mode with 1/4 window option. The two gain calibrations are consistent to within $\sim$5eV over the entire energy band for the whole mission to date.

Figure 7.14: Comparison of the gain in the Normal mode with 1/4 window option (red) and with no option (black) for five conspicuous emission lines in XIS0. Events of all segments are used.

Figure 7.15: Comparison of the gain in the Normal mode with 1/4 window option (red) and with no option (black) for five conspicuous emission lines in XIS1. Events of all segments are used.

Figure 7.16: Comparison of the gain in the Normal mode with 1/4 window option (red) and with no option (black) for five conspicuous emission lines in XIS3. Events of all segments are used.
[XIS3]\includegraphics[width=0.9\textwidth]{figures_xis/1_4_win_xis3} Normal mode (burst option)

For the calibration of the Normal mode with burst option advantage is taken of the Crab calibration observations (which are primarily performed for HXD calibration purposes). Since the Crab has a featureless spectrum, the $^{55}$Fe calibration source spectra are used in addition (Table 7.10). In the Normal mode with burst option no correction is made in addition to those for the Normal mode without options. The calibration observations are only used to confirm that there is no change in the instrumental response for the burst options. Fig. 7.17 shows a $^{55}$Fe calibration source spectrum taken with the 2.0s burst option during an observation of GX 17$+$2. It has been modeled using the RMF for the Normal mode (without options). There is no structure in the residuals. This is indicating that the response is the same for the Normal mode without options and the Normal mode with the 2.0s burst option.

Figure 7.17: $^{55}$Fe spectrum of a 2.0s burst mode observation of GX17$+$2. The spectrum is fitted with a continuum plus two Gaussian lines using the response generated for the Normal mode (no burst option).
\includegraphics[width=0.75\textwidth,angle=270]{figures_xis/xis1_segA} P-sum mode

Because of the unavailability of the spaced-row charge injection technique for the P-sum clocking mode (§ 7.3.1), the data taken in this mode suffer from a substantially worse energy resolution than those taken with the Normal clocking mode. The P-sum clocking mode is only combined with the timing editing mode (§, which has limited information about the charge distribution around an event. This makes the background level of the P-sum mode much higher than that of the Normal mode. Fig. 7.18 shows the history of phenomenological gain of the P-sum data. Fig. 7.19 shows the energy resolution for two representative lines, one in the soft and one in the hard band.

Figure 7.18: History of the P-sum gain.

Figure 7.19: History of the P-sum energy resolution. The upper panel shows the change for the Ne IX K$\alpha $ line, whereas the lower panel shows the change for the Mn I K$\alpha $ line.
\includegraphics[width=0.65\textwidth,angle=270]{figures_xis/FWHM_history_XIS0-3_NeIX} \includegraphics[width=0.65\textwidth,angle=270]{figures_xis/FWHM_history_XIS0-3_Mnka} Energy Calibration Results (3) -- Other Editing Modes 2$\times $2 editing mode

The data taken with the 2$\times $2 editing mode have insufficient information to conduct trail correction around each event, unlike those taken with the 3$\times $3 or 5$\times $5 editing modes. It is therefore inevitable for the 2$\times $2 editing mode to have different gain compared to the other editing modes. In particular, the gain difference between 3$\times $3/5$\times $5 and 2$\times $2 depends on the ACTY position of the imaging area. Since the 2$\times $2 editing mode is only used for very bright point-like sources for the purpose of reducing telemetry and since these observations are made at the XIS nominal position, the 2$\times $2 editing mode data are calibrated such that the gain difference between 3$\times $3/5$\times $5 and 2$\times $2 modes is minimal at the image center. Fig. 7.20 shows the difference of the 2$\times $2 and 5$\times $5 editing mode gains in the soft and hard energy bands at several different ACTY positions. At the center (ACTY$=$512), the difference is within $\sim$3eV. For calibration purposes, 2$\times $2 editing mode data are artificially generated on the ground using the 3$\times $3 or 5$\times $5 editing mode data for the Perseus cluster (Table 7.10).

Figure 7.20: Energy gain for the 2$\times $2 and 5$\times $5 editing modes as a function of ACTY position for XIS0 and XIS3 in the soft and hard energy band for several different epochs: 2008/02 (black), 2009/02 (red), 2010/02 (green), and 2011/02 (blue).
[XIS0, soft-band]\includegraphics[width=0.45\textwidth]{figures_xis/2x2a} [XIS3, soft-band]\includegraphics[width=0.45\textwidth]{figures_xis/2x2b} [XIS0, hard-band]\includegraphics[width=0.45\textwidth]{figures_xis/2x2c} [XIS3, hard-band]\includegraphics[width=0.45\textwidth]{figures_xis/2x2d} Energy Calibration Notes for Users & Proposers

Users can exploit the calibration results presented above by using the xisrmfgen tool. Remaining issues include the following:

Si K
In all clocking/editing modes, uncertainties of the gain and energy resolution calibration at the Si K edge remain prominent for the FI devices. It is currently advised to remove data between 1.8 and 2.0keV from spectral fitting. For the BI device, a workaround has been implemented in the CALDB in 2012 which is mitigating this uncertainty.
Au M
Some uncertainties still remain in the Au M band, with Au having been used for the surface coating of the mirrors. This is visible for very bright sources. Users need to remove data taken in the relevant energy band from spectral fitting.
Al K
Some uncertainties still remain in the Al K band, with Al having been used for the electrodes in the CCD. This is visible for very bright sources. Users need to remove data taken in the relevant energy band from spectral fitting.
The background level is higher than expected in the energy band above $\sim$10keV. The cause for this is not yet identified.

7.3.3 Contamination & Low Energy Quantum Efficiency Degradation

The quantum efficiency below $\sim$2keV has been decreasing since launch due to accumulation of contaminating material on the optical blocking filter (OBF) of each sensor. The OBF is cooler than other parts of the satellite, thus it is prone to the accumulation of contamination. The contaminant consists of several different materials with time-varying composition. The time dependence of the thickness and the chemical composition of the contaminant at the XIS nominal position are monitored and calibrated, as well as the spatial dependence of the thickness across the field of view (Table 7.10). Contamination Calibration Results Time dependence of the chemical composition of the contaminant

The chemical composition is modeled phenomenologically with time-varying columns of H, C, N and O. Three calibration sources -- RXJ1856.5$-$3754 (a super-soft isolated neutron star), PKS2155$-$304 (a blazar), and 1E0102.2$-$7219 (a line-dominated supernova remnant) -- are used to derive the chemical composition at each epoch of the observations. The spectral models by Burwitz et al. (2003) for RXJ1856.5$-$3754, a simple power-law model for PKS2155$-$304, and the model by Plucinsky et al. (2008) for 1E0102.2$-$7219 are used as standard models. Fig. 7.21 shows the result of XIS1 fitting for RXJ1856.5$-$3754 using the 2012 contamination model for selected epochs.

Figure 7.21: Results of spectral fitting of the BI spectrum of RXJ1856.5$-$3754 with the 2012 contamination model at the XIS nominal position for several different epochs.
\includegraphics[height=0.55\textheight,angle=270]{figures_xis/xissimarfgen_hcno_fixnorm} Time dependence of the thickness of the contaminant

The time dependence of the thickness of the contaminant is modeled with phenomenological functions of time, separately for each composition and sensor. Fig. 7.22, 7.23, and 7.24 show the evolution of the thickness for different elements, while Fig. 7.25 shows the combined optical thickness and the relative reduction of the effective area for two different energies. The thickness increased rapidly for one year after launch and continued to increase at a moderate pace thereafter. The N component is used for the XIS1 only.

Figure 7.22: Time dependence of the C thickness at the XIS nominal position for each sensor. The solid curves indicate the phenomenological trend model.

Figure 7.23: Time dependence of the N thickness at the XIS nominal position for each sensor. The solid curves indicate the phenomenological trend model.

Figure 7.24: Time dependence of the O thickness at the XIS nominal position for each sensor. The solid curves indicate the phenomenological trend model.

Figure 7.25: Time dependence of the combined optical depth (top panel) and the relative reduction of the effective area (bottom panel). In both panels circles are used for measurements at 0.65keV and crosses for those at 1.0keV.
\includegraphics[height=0.40\textheight]{figures_xis/tau_comb_hcno_20120420} Spatial dependence of the thickness of the contaminant

The spatial distribution of the contaminants is monitored and calibrated using regular observations of a part of the Cygnus Loop, a thermal supernova remnant. The atmospheric fluorescent K lines of N I and O I, which illuminate the entire field of view when the telescope is oriented toward the day earth, are used as well. The model of the spatial dependence assumes a radially symmetric pattern (Koyama et al., 2007). The chemical composition and the thickness at the center are normalized to the values presented in Fig. 7.22, 7.23, and 7.24. Different sensors have different radial profiles (Fig. 7.26).

Figure 7.26: Spatial dependence of the thickness derived from day earth observation data at different epochs. Open circles and filled triangles represent the data points determined by the N and O lines, respectively. The best fit models are shown as solid curves.
\includegraphics[height=0.4\textheight]{figures_xis/Contami_Dist} Contamination Notes for Users & Proposers

The contaminant calibration results are available for use through the ARF generation tools xissimarfgen (Ishisaki et al., 2007) and xisarfgen. It is discouraged to use the generic contamination models in the XSPEC package (e.g., xisabs, xiscoabs, xiscoabh, xispcoab). For proposal writers ARF files are provided that assume a contamination thickness extrapolated to the middle of the next AO cycle, based on the current trend. Although a major revision of the contamination model was made in 2012, that model still shows some discrepancies compared to observations. Fig. 7.27 shows results of fitting E0102$-$72 data using the 2012 contamination model.

Figure 7.27: Results of fitting the 2012 contamination model using the E0102$-$72 data taken on 2011-06-29.
[XIS0]\includegraphics[width=0.48\textwidth]{figures_xis/caldb_xis0_20110629} [XIS1]\includegraphics[width=0.48\textwidth]{figures_xis/caldb_xis1_20110629} [XIS3]\includegraphics[width=0.48\textwidth]{figures_xis/caldb_xis3_20110629}

7.3.4 Non X-Ray Background

The Non X-ray Background (NXB) has several sources. The most dominant source are events produced by ionization losses of charged cosmic ray particles. Others include fluorescence X-rays from materials used in the spacecraft and the $^{55}$Fe calibration sources attached to the door of each sensor (opened immediately after the launch). Most of the NXB events are discarded onboard as grade 7 events. The NXB of the XIS is known to be very stable on time scales of months and thus the NXB spectrum can be constructed using data obtained when the spacecraft is pointed toward the Earth at night. The NXB database is accessible as part of the CALDB, which is updated biannually in June and December. NXB Calibration Results NXB Intensities and spectra

Fig. 7.28 shows the NXB spectrum for each sensor. The background rate in the 0.4-12keV band is 0.1-0.2counts s$^{-1}$ for the FI CCDs and 0.3-0.6counts s$^{-1}$ for the BI CCD after grade selection. Table 7.11 shows the current best estimates for the strength of major XIS emission features, along with their 90% confidence errors.

Figure 7.28: Non X-ray background (NXB) rate for XIS0 (black), XIS1 (red), and XIS3 (blue). The spectra are constructed from night Earth observations.

Table 7.11: Strength of conspicuous emission lines in the NXB spectrum. The count rates are obtained from the entire CCD chip except for the corners irradiated by the calibration source. The errors are the 90% statistical uncertainties (Tawa et al., 2008).
Line Energy XIS0 XIS1 XIS2 XIS3
  [keV] [$10^{-9}$cps/pix] [$10^{-9}$cps/pix] [$10^{-9}$cps/pix] [$10^{-9}$cps/pix]
Al K$\alpha $ 1.486 $1.45\pm0.11$ $1.84\pm0.14$ $1.41\pm0.10$ $1.41\pm0.10$
Si K$\alpha $ 1.740 $0.479\pm0.081$ $2.27\pm0.15$ $0.476\pm0.080$ $0.497\pm0.082$
Au M$\alpha $ 2.123 $0.63\pm0.093$ $1.10\pm0.13$ $0.776\pm0.097$ $0.619\pm0.092$
Mn K$\alpha $ 5.895 $6.92\pm0.19$ $0.43\pm0.14$ $1.19\pm0.13$ $0.76\pm0.11$
Mn K$\beta$ 6.490 $1.10\pm0.11$ $0.26\pm0.13$ $0.40\pm0.11$ $0.253\pm0.094$
Ni K$\alpha $ 7.470 $7.12\pm0.19$ $7.06\pm0.37$ $8.01\pm0.20$ $7.50\pm0.20$
Ni K$\beta$ 8.265 $0.96\pm0.10$ $0.75\pm0.22$ $1.16\pm0.11$ $1.18\pm0.11$
Au L$\alpha $ 9.671 $3.42\pm0.15$ $4.15\pm0.49$ $3.45\pm0.15$ $3.30\pm0.15$
Au L$\beta$ 11.51 $2.04\pm0.14$ $1.93\pm0.48$ $1.97\pm0.14$ $1.83\pm0.14$ NXB cut-off rigidity dependence

The total intensity of the NXB depends strongly on the geomagnetic cut-off rigidity (COR), as the dominant source of the background originates from cosmic rays. Fig. 7.29 shows the NXB count rate as a function of the COR. The ftool xisnxbgen generates NXB spectra for an observation in such a way that the histogram of the COR is the same between the observation of the source and the night Earth observations. The night Earth data are retrieved from $\pm $150 days of the observation of the source by default. The PIN's Upper Discriminator (PIN-UD) count rate is also useful as a proxy for the COR, and thus the XIS NXB level. Tawa et al. (2008) show that the PIN-UD provides a slightly better reproducibility of the XIS NXB than the COR. The reproducibility of the NXB in the 5-12keV band is evaluated to be 3-4% of the NXB, when the PIN-UD is used as the NXB sorting parameter.

Figure 7.29: Cut-off rigidity dependence of the NXB (average intensity for 5-10keV) for each sensor. The NXB flux varies by a factor of $\sim$2 depending on the cut-off rigidity.
\includegraphics[totalheight=4in]{figures_xis/fig11} NXB spatial dependence

The NXB is not uniform over the chip. It is stronger toward larger ACTY positions (Fig. 7.30). This is because some fraction of the NXB is produced in the frame-store region. The fraction can be different between the fluorescent lines and the continuum.

Figure 7.30: ACTY dependence of the NXB for XIS0, XIS1, and XIS3. Black lines indicate the continuum component (2.5-5.5keV), while the red lines indicate the Ni K$\alpha $ line component (7.2-7.8keV). The NXB flux tends to be higher at larger ACTY, because some fraction of the NXB is produced in the frame store region.
\includegraphics[height=2.5in]{figures_xis/fig10_xis0} \includegraphics[height=2.5in]{figures_xis/fig10_xis1} \includegraphics[height=2.5in]{figures_xis/fig10_xis3} \includegraphics[height=2.5in]{figures_xis/fig10_dummy} NXB time dependence

The NXB level changes both continuously and discontinuously. The continuous changes are seen only in the low energy band for the BI sensor, in which a gradual increase in the NXB level is observed (Figure 7.31). The level is stable for the high energy band for the BI and the total band for the FI sensors.

Figure 7.31: Soft NXB spectrum of XIS1 (0.1-0.6 keV) integrated over one year for several epochs. For the CI$=$6keV data, the second trailing rows are not removed.
\includegraphics[width=0.6\textwidth]{figures_xis/nxb_xis1_time} NXB Notes for Users & Proposers NXB change due to the XIS0 anomaly

A putative micro-meteorite hit occurred for XIS0 in 2009-06-23. Since then, the XIS0 has been operated with an area discriminator masking the damaged area. In the masked area, the NXB level is zero, which causes an apparent discontinuous change in the NXB database. Users generating their own NXB spectrum using xisnxbgen need to be aware that they are mixing the NXB data before and after the event if their observations are within $\pm $150 days of the event (between January 24, 2009 and June 27, 2009). A recipe for mitigating this is described at NXB change due to the XIS1 SCI increase

The SCI level was increased from 2 to 6keV in 2010. Due to the increased amount of charges, the NXB level has increased discontinuously. The increase is only seen in the second trailing rows, which are the over-next rows from the rows with charge injection. Users can achieve the same NXB level as for SCI$=$2keV by masking events from the second trailing row1s. Details can be found at NXB contamination due to the day Earth

When the XIS field of view is close to the day Earth (i.e., Sun-lit Earth), fluorescent lines from the atmosphere contaminate the low energy part of the XIS data, especially for the BI chip. Most prominent are the O and N lines. Although the standard event screening criterion (elevation angle from the day Earth $>$20degrees) is sufficient to remove these features, some emission may remain due to the variable nature of the Earth's X-ray albedo. In this case, event screening with a higher elevation angle is recommended.

7.3.5 Flux Calibration

The stability of the relative normalization between the three XIS sensors is shown in Fig. 7.32. No significant change with time is found. The relative normalization remains constant. The mean and standard deviation are summarized in Table 7.12 separately for the XIS and HXD nominal positions.

Figure 7.32: Relative flux normalization among different sensors. The ratio of the best-fit normalization values are compared between different combinations of two sensors, at the XIS and the HXD nominal positions. Observations of a power-law source in the Normal clocking mode were used. The average values are shown with solid lines.

Table 7.12: Relative normalization among different sensors.
Position Ratio Mean Standard deviation
XIS-norm XIS1/XIS0 1.026 0.016
  XIS3/XIS0 1.014 0.017
HXD-norm XIS1/XIS0 1.004 0.019
  XIS3/XIS0 0.980 0.022

7.3.6 Putative Micro-Meteorite Hits

Sudden anomalies caused putatively by micro-meteorite hits have affected XIS0, XIS1, and XIS2. The entire XIS2 was lost in 2006. A part of the XIS0 was lost in 2009, which is masked by area discrimination since then. A very small hole was created in the optical blocking filter for the XIS1, but the sensor remains intact. Event on 2006-11-09 for XIS2

The anomaly of XIS2 suddenly occurred on Nov. 9, 2006, 1:03 UT. About 2/3 of the image was flooded with a large amount of charge, which had leaked somewhere in the imaging region. When the anomaly occurred, the satellite was out of the SAA and the XIS sensors were conducting observations in the Normal mode (SCI on, without any options). Various tests to check the condition of XIS2 revealed that (1) the four readout nodes of the CCD and the corresponding analog chain were all working fine, (2) the charge injection was not directly related to the anomaly, but may have helped to spread the leaked charge. When the clock voltages were changed in the imaging region, the amount of leaked charge changed as well. This indicates a short between the electrodes and the buried channel. Possible mechanisms to cause the short include a micro-meteoroite impact on the CCD, as seen, e.g., on XMM-Newton and Swift. Although there is no direct evidence to indicate an micro-meteoroite impact, the phenomenon observed for XIS2 is not very different from what is expected from such an event. The low-earth orbit and the low-grazing-angle mirrors of Suzaku may have enhanced the probability of a micro-meteoroite impact. An attempt to reduce the leaked charge by changing the clock pattern and voltages in the imaging region was not successful. Therefore it was decided to stop operating the sensor. It is unlikely that operation of the XIS2 will be resumed in the future. More details can be found at Event on 2009-06-23 for XIS0

XIS0 suddenly showed an anomaly on June 23, 2:00 UT. During Normal clocking operations, a part of segment A of XIS0 was flooded with a large amount of charges, which caused saturation of the analogue electronics. The anomaly was very similar to that which occurred in the XIS2 in 2007. It is therefore suspected that both anomalies have the same origin, possibly a micro-meteorite impact.The effect is confined to 1/8 of the area of XIS0. The XIS team continues to operate XIS0. Users need to be aware of several remaining artifacts after the event:. In the Psum clocking mode, the effect spreads to the entire XIS0 with severe data degradation. The XIS team discontinues the use of Psum clocking mode for the XIS0. More details can be found at Event on 2009-12-18 for XIS1

XIS1 suddenly showed an anomaly some time between Dec 18, 2009 12:50 UT and 14:10 UT. A bright and persistent spot suddenly appeared at one end of segment C in all images taken during day Earth observations, while none was found during night Earth observations. It is speculated that the anomaly stems from optical light leaked from a hole with a size of $\sim$7.5$\mu $m in the optical blocking filter created by a micro-meteorite hit. From diagnostic observations, it was concluded that the scientific impact of this anomaly is minimal. XIS1 has been and will be operated in the same way as before the anomaly. More details can be found at

7.4 XIS Onboard Processing

7.4.1 Pulse Height Determination & Hot Pixel Rejection

When a CCD pixel absorbs an X-ray photon, the X-ray is converted to an electric charge, which in turn produces a voltage at the analog output of the CCD. This voltage (``pulse height'') is proportional to the energy of the incident X-ray. In order to determine the true pulse height corresponding to the input X-ray energy, it is necessary to subtract the dark levels and correct possible optical light leaks.

Dark levels are non-zero pixel pulse heights caused by leakage currents in the CCD. In addition, optical and UV light might enter the sensor due to imperfect shielding (``light leak''), producing pulse heights that are not related to X-rays. The analysis of ASCA SIS data, which utilized an X-ray CCD in photon-counting mode for the first time, showed that the dark levels were different from pixel to pixel, and the distribution of the dark level did not necessarily follow a Gaussian function. On the other hand, light leaks are considered to be rather uniform over the CCD.

For the Suzaku XIS, the dark levels and the light leaks are calculated separately in the Normal mode. The dark levels are defined for each pixel and are expected to be constant for a given observation. The PPU calculates the dark levels in the Dark Initial operation; those are stored in the Dark Level RAM. The average dark level is determined for each pixel, and if the dark level is higher than the hot-pixel threshold, this pixel is labeled as a hot pixel. The dark levels can be updated by the Dark Update operation, and sent to the telemetry by the Dark Frame mode. The analysis of ASCA data showed that the dark levels tend to change mostly during the SAA passage of the satellite. The Dark Update operation is conducted after every SAA passages unless the telescope is pointed toward the day Earth.

Hot pixels are pixels which always output pulse heights larger than the hot-pixel threshold even without input signals. Hot pixels are not usable for observations, and their output has to be disregarded during scientific analysis. The XIS detects hot pixels on-board by the Dark Initial/Update mode, and their positions are registered in the Dark Level RAM. Thus, hot pixels can be recognized on-board, and they are excluded from the event detection processes. It is also possible to specify hot pixels manually. However, some pixels output pulse heights larger than the threshold intermittently. Such pixels are called flickering pixels. It is difficult to identify and remove flickering pixels on board. They are inevitably included in the telemetry and need to be removed in scientific analysis, for example by using the FTOOLS sisclean. Flickering pixels sometimes cluster around specific columns, which makes them relatively easy to identify.

The light leaks are calculated on board from the pulse height data after subtraction of the dark levels. A truncated average is calculated for $256\times114$ pixels (this size was $64\times64$ before January 18, 2006) in every exposure and its running average produces the light leak. In spite of the name, light leaks do not represent in reality optical/UV light leaks in the CCD. They mostly represent fluctuation of the CCD output correlated to the variations of the satellite bus voltage. The XIS has little optical/UV light leak,it is negligible unless the bright earth comes close to the XIS field of view.

The dark levels and the light leaks are merged in the Parallel-sum (P-Sum) mode, so the Dark Update mode is not available in the P-Sum mode. The dark levels, which are defined for each pixel as in the case of the Normal mode, are updated every exposure. It may be considered that the light leak is defined for each pixel in the P-Sum mode.

7.4.2 On-Board Event Analysis

The main purpose of the on-board processing of the CCD data is to reduce the total amount of data transmitted to the ground. For this purpose, the PPU searches for a characteristic pattern of the charge distribution (called an event) in the pre-processed (post dark level and light leak subtraction) frame data. When an X-ray photon is absorbed in a pixel, the photo-ionized electrons can spread into at most four adjacent pixels.

An event is recognized when a pixel has a pulse height which is between the lower and the upper event thresholds and is larger than those of eight adjacent pixels (e.g., it is the peak value in the $3\times3$ pixel grid). In the P-Sum mode, only the horizontally adjacent pixels are considered. The copied and the dummy pixels ensure that the event search is enabled on the pixels at the edges of each segment. The RAW XY coordinates of the central pixel are considered the location of the event. Pulse-height data for the adjacent $5
\times 5$ square pixels (or three horizontal pixels in the P-Sum mode) are sent to the Event RAM as well as the pixel location.

The MPU reads the Event RAM and edits the data to the telemetry format. The amount of information sent to the telemetry depends on the editing mode of the XIS. All the editing modes are designed to send the pulse heights of at least four pixels of an event to the telemetry, because the charge cloud produced by an X-ray photon can spread into at most four pixels. Information of the surrounding pixels may or may not be output to the telemetry depending on the editing mode. The $5
\times 5$ mode outputs the most detailed information to the telemetry, i.e., all 25 pulse-heights from the $5
\times 5$ pixels containing the event. The size of the telemetry data per event is reduced by a factor of two in the $3\times3$ mode, and another factor of two in the $2\times2$ mode. Details of the pulse height information sent to the telemetry are described in the next section.

7.4.3 Discriminators

Three kinds of discriminators, area, grade, and class discriminators, can be applied during the on-board processing. The grade discriminator is available only in the Timing mode. The class discriminator was implemented after the launch of Suzaku and was used since January, 2006. In most cases, guest observers need not change the default setting of these discriminators.

The class discriminator classifies the events into two classes, ``X-rays'' and ``others,'' and outputs only the "X-ray" class to the telemetry when it is enabled. This class discriminator is always enabled to reduce the telemetry usage of non-X-ray events. The ``other'' class is close to, but slightly different from the ASCA grade 7. When the XIS points to a blank sky, more than 90% of the detected events are particle events (mostly corresponding to ASCA grade 7 events). If we reject these particle events on board, we can make a substantial saving in the telemetry usage. This is especially useful when the data rate is medium or low. The class discriminator realizes such a function in a simple manner. When all the eight pixels surrounding the event center exceed the Inner Split Threshold, the event is classified as from the ``other'' class, and the rest of the events as from the ``X-ray'' class. With such a simple method, we can reject more than three quarters of the particle events. The class discriminator works only for the $5
\times 5$ and $3\times3$ modes. It is not available in the $2\times2$ mode and the timing mode.

Fig. 7.5 shows the pixel pattern for which the pulse height or 1-bit information is sent to the telemetry. We do not assign grades to an event on board in the Normal Clock mode. This means that a dark frame error, if present, can be corrected accurately during ground processing even in the 2 $\times $ 2 mode. The definition of the grades in the P-Sum mode is shown in Fig. 7.33.

Figure 7.33: Definition of grades in the P-Sum/Timing mode. Total pulse height and the grade of the event are output to the telemetry. Note that the grades are defined referring to the direction of the serial transfer, so the central pixel of the grade 1 event has a larger RAW-X value than the second pixel, while the opposite is true for the grade 2 event.

next up previous contents
Next: Bibliography Up: Suzaku Technical Description Previous: 6. X-Ray Telescopes (XRT)   Contents
Michael Arida 2012-10-19