4.1.

Soft X-ray Telescopes

The SXT^21–25 is very similar to the Suzaku X-ray Telescope²⁶ but with a longer focal length of 5.6 m and a larger outer diameter of 45 cm. The SXT consists of three parts: an x-ray mirror, a stray light baffle called the pre-collimator, and a thermal shield to keep the mirror temperature at around 20°C. The mirror is a conically approximated Wolter I grazing incidence optic with 203 nested shells. Each shell is segmented into four quadrants.

The flight SXT mirror assemblies [Fig. 4(a)], SXT-I for the SXI and SXT-S for the SXS, were fabricated at NASA/GSFC and delivered to JAXA. According to calibration at GSFC and ISAS, the angular resolution [half-power diameter (HPD)] is 1.3 and 1.2 arc min for the SXT-I and SXT-S, respectively. The result obtained with SXT-S exceeds the desired goal. Effective areas were measured to be $\sim 590{\mathrm{cm}}^2$ at 1 keV and $\sim 430{\mathrm{cm}}^2$ at 6 keV. The system net effective area at 1 and 6 keV is about 250 and $300{\mathrm{cm}}^2$ for the SXS and 370 and $350{\mathrm{cm}}^2$ for the SXI, respectively. However, since the SXS gate valve (GV) was closed during observations, the SXS effective area was reduced by the Be window transmission and the GV obscuration. The SXS net effective area with the GV closed was about $160{\mathrm{cm}}^2$ at 6 keV with almost no area below 2 keV.

Fig. 4

Photographs of flight models of (a) Soft X-ray Telescope, SXT-S, and (b) Hard X-ray Telescope, HXT-1.

According to the in-flight data, the SXT-I and SXT-S were clearly focusing x-rays onto the corresponding detectors and they seemed to be working as expected. The overall effective area (flux) was examined using the Crab and the G21.5-0.9 data. Model fittings for both spectra produced a power-law photon index consistent with the previous measurements,^27,28 which indicates that the effective area response was consistent with the ground calibration results.²⁹ The measured HPD was 1.3 and 1.2 arc min for SXT-I and SXT-S, respectively. Detailed in-flight performance of the SXT is presented in other publications.³⁰

4.2.

Hard X-ray Telescopes

A depth-graded multilayer mirror reflects x-rays not only by total external reflection but also by Bragg reflection. To obtain a high reflectivity up to 80 keV, the HXT consists of a stack of multilayers with different sets of periodic length and number of layer pairs with a platinum/carbon coating. The technology of a hard x-ray focusing mirror has already been proven by the balloon programs $\mathrm{InFOC}\mu \mathrm{S}$ (2001, 2004),^31,32 HEFT (2004),³³ SUMIT (2006),³¹ and recently with the NuSTAR satellite.³⁴

The HXT^35–40 consists of three parts: an x-ray mirror, a stray light baffle (pre-collimator), and a thermal shield [Fig. 4(b)]. The mirror is based on conically approximated Wolter I grazing incidence optics.^37,38 The diameters of the innermost and the outermost reflectors are 120 and 450 mm, respectively. The total number of nested shells is 213. Since a telescope module is made up from three segments with an azimuthal opening angle of 120 deg each, it requires 1278 reflectors in total. Production of two flight-ready HXT mirror assemblies, HXT-1 and HXT-2, was completed in 2014. According to calibration performed using the beam line BL20B2 of the synchrotron radiation facility SPring-8, the characteristics of HXT-1 and HXT-2 are quite similar. Based on the ground calibration at the SPring-8, a collecting area of $174{\mathrm{cm}}^2$ at 30 keV for one telescope has been achieved, resulting in a total effective area of $348{\mathrm{cm}}^2$. The HPD of the HXTs is $\sim 1.9\mathrm{arc}\min$ at 30 keV.^40,41 After the launch, the characteristics of HXTs were evaluated using observational data of the Crab Nebula. The HPDs of HXT-1 and HXT-2 in the 5- to 80-keV band were 1.59 and 1.65 arc min, respectively. Note, however, that the encircled energy function (EEF) was normalized at $r=6\mathrm{arc}\min$ on the ground whereas, due to the limitation of the HXI field of view, the in-flight EEF was normalized at $r=4\mathrm{arc}\min$ and the data in the area $r>4\mathrm{arc}\min$ were subtracted as background. We, therefore, reprocessed the ground calibration data in the same way as the in-flight data. The reanalyzed HPDs at 30 keV were 1.77 and 1.84 arc min for HXT-1 and HXT-2, respectively. The in-flight HPDs seem to be slightly smaller than the ground ones. However, both the in-flight and ground HPDs can have an uncertainty of 0.1 arc min. Thus, they are consistent with each other within the uncertainty. The unabsorbed flux of the Crab nebula in the 3- to 50-keV band was measured to be $3.59\times {10}^{-8}\mathrm{erg}{\mathrm{s}}^{-1}{\mathrm{cm}}^{-2}$ with HXT-1 and $3.55\times {10}^{-8}\mathrm{erg}{\mathrm{s}}^{-1}{\mathrm{cm}}^{-2}$ with HXT-2. These values are slightly ($\sim 5\%$) larger than that measured with NuSTAR. Detailed in-flight performance of the HXT is presented in other publications.⁴²

4.3.

Soft X-ray Spectrometer System

The SXS system consists of the SXT, the filter wheel (FW) assembly,⁴³ and the SXS.^44–47 The SXS is a 36-pixel system with an energy resolution of better than 7 eV between 0.3 and 12 keV. The array design for the SXS is basically the same as that for the Suzaku XRS⁴⁸ but has larger pixel pitch and absorber size. HgTe absorbers are attached to ion-implanted Si thermistors formed on suspended Si microbeams.^49–51 The $6\times 6\text{array}$ of silicon thermistors on an $832\text{-}\mu \mathrm{m}$ pitch was manufactured during the Suzaku/XRS program along with arrays with smaller pitch as an option for a larger field of view [Fig. 5(a)]. For SXS, improved heat sinking was added to the frame of the array, and HgTe absorbers with very low specific heat were attached to the pixels. The width of the individual absorbers was $819\mu \mathrm{m}$.⁵¹ A square-cm anticoincidence detector was installed behind the microcalorimeter array in the detector assembly. The detector assembly consisted of the detector enclosure and its mechanical, thermal, and electrical interfaces, including the first stage of amplification.

Fig. 5

Photographs of (a) SXS sensor and (b) SXS dewar. The sensor was suspended from the outer structure using Kevlar, and electrical connections to the housing were made using tensioned wires to reduce the sensitivity to microphonics. The outer shell of the dewar is 950 mm in diameter.

The SXS detector assembly and low-temperature coolers were installed in the dewar [Fig. 5(b)]. The cooling system was required to cool the array to 50 mK with sufficient duty cycle to fulfill the SXS scientific objectives. This restriction requires extremely low heat loads. To achieve the necessary gain stability and energy resolution, the cooling system must regulate the detector temperature, to within $2\text{-}\mu \mathrm{K}$ rms over intervals of about half an hour, for at least $24\mathrm{h}/\text{cycle}$.⁵² From the detector stage to room temperature, the cooling chain is composed of a three-stage adiabatic demagnetization refrigerator (ADR),⁵³ superfluid liquid ${}^4\mathrm{He}$ (hereafter LHe), a ${}^4\mathrm{He}$ Joule–Thomson (JT) cryocooler, and two-stage Stirling cryocoolers.

To obtain a good performance for bright sources, an FW assembly, which includes a wheel with selectable filters and a set of modulated x-ray sources, were provided by SRON and the University of Geneva. This was placed at a distance of 90 cm from the detector. The FW is able to rotate a suitable filter into the beam to optimize the quality of the data, depending on the source characteristics.⁴³ In addition to the filters, a set of on–off-switchable x-ray calibration sources, using a light-sensitive photocathode, were available. With these calibration sources, the energy scale could be calibrated with a typical 1- to 2-eV accuracy, allowing proper gain and linearity calibration of the detector in-flight.

All SXS components except for the SXS power distributor (SXS-DIST) were powered off at launch (cold launch). The first critical operation was to re-establish the He plumbing in space.⁵⁴ Quickly after the fairing opened, the He vent valve was opened, which was confirmed by the telemetry during the first contact 50 min after the launch. We next focused on starting the cooling chain. On day 0, we started two shield coolers at the full wattage and two precoolers (PC) at a low wattage. On day 1, the PCs were also ramped up to the full wattage. By day 4, the JT cooler was also ramped up step by step. On day 5, the ADR cool down was started and the sensor was cooled to 50 mK. After that, the cooling chain worked without any problem until the end of the mission by repeating ADR recycles periodically and continuously operating the cryocoolers.

The signal chain was also started within the first week. On day 2, both the analog and digital signal processors were started (called XBOX and PSP, respectively). After the initial check and the parameter setting, the first noise measurement was performed on day 2 at a warm detector temperature and on day 5 at the 50-mK temperature.

The SXS was ready for observations by as early as day 6. On day 7, the spacecraft was pointed to the Perseus cluster, and we observed x-rays from an astronomical target for the first time. Although the pointing was offset from the cluster core by a few arc minutes, we obtained sufficient counts to make an offset correction based on our own count rate map within a few cycles. On day 8, an adjustment maneuver was made toward the core of the cluster for a longer exposure time. The detector NXB level measured during the Earth occultation was very low ($<1.0\times {10}^{-3}\mathrm{cts}{\mathrm{s}}^{-1}{\mathrm{keV}}^{-1}$).

After the EOB extension, we started commissioning the FW electronics, which was the last SXS subsystem to be powered. The FW was rotated on day 30 to provide x-ray illumination of the whole array by ${}^{55}\mathrm{Fe}$ sources. The resulting calibration data set was used to refine the gain scales and measure the energy resolutions of the individual pixels. Figure 6(a) is a histogram of the distribution of resolutions measured across the array. The composite resolution of the whole array was 4.9 eV (FWHM) [Fig. 6(b)].^51,55,56

Fig. 6

(a) A histogram of the distribution of resolutions measured across the array. (b) The whole-array spectrum around 5.9 keV. The points are the data, with $\mathrm{sqrt}(N)$ error bars on the counts barely visible. The dark line is the best fit to a model of the natural line shape convolved with Gaussian broadening, and the dashed gray line is the natural line shape.

What remained to be done during the commissioning phase but was not completed were: (i) the start-up of the modulated x-ray source, (ii) opening of the GV and the start of the temperature control of the dewar main shell filter, and (iii) the final setting of the event threshold. The cryogen-free operation using the third stage ADR, which was planned in later phase of the mission,⁵⁷ was also not carried out.

4.4.

Soft X-ray Imager

X-ray sensitive silicon CCDs are key detectors for x-ray astronomy. The low background and high-energy resolution combined with wide field of view achieved with the Suzaku XIS clearly shows that the x-ray CCD can also play a very important role in the mission. The soft x-ray imaging system consists of an x-ray mirror, the SXT-I, and a CCD camera, the SXI.^58–63,64 Figure 7(a) shows a photograph of the SXI detector. The SXI camera contains a cold plate on which four CCDs are placed. The cold plate is connected to two identical single-stage Stirling coolers.

Fig. 7

Photographs of (a) SXI, (b) HXI, and (c) SGD.

Start-up operation of SXI began from March 2, 2016. The CCD temperature reached the nominal value, $-110°\mathrm{C}$, on March 7, and we started data acquisition in the event mode. One of the Stirling coolers was used to cool down the CCDs. Fine control of the temperature was performed with heaters attached to the cold plate, and we confirmed the same level of temperature stability as that during the ground tests. We employed a charge injection technique from the beginning of the operation to cope with the decrease of the charge transfer efficiency due to radiation damage. Artificial charge was injected from the top of columns of the CCDs at every 160 rows (before the on-chip $2\times 2$ binning). We confirmed that the noise level of CCDs was about 5 to $7{\mathrm{e}}^{-}\mathrm{rms}$, the same as that of the ground tests. We also confirmed that the amount of the injected charge was not changed from the ground test.

The first target was the Perseus cluster of galaxies. As shown in Fig. 8(a), the cluster image is offset from the SXI center, because the CCDs were placed to have an offset relative to the aim point of SXT-I by 5 arc min to avoid the CCD gaps. Two calibration sources of ${}^{55}\mathrm{Fe}$ are also seen as semicircular shapes on top and bottom centers. Initial performance of the SXI was checked using these calibration source data and the Perseus data. The gain and the energy resolution of the CCDs were found to be roughly consistent with the ground data.⁶⁵ Thus, we confirmed that the SXI functioned properly on orbit. We accumulated background data from the source-free regions of the SXI field of view. In the case of Suzaku XIS(BI),⁶⁶ non-x-ray background NXB increased rapidly above 6 keV. Such an increase of NXB almost disappeared for the SXI due to the thick depletion layer ($200\mu \mathrm{m}$) compared to that of the XIS(BI) ($\sim 42\mu \mathrm{m}$). Furthermore, the Ni K lines in the NXB spectrum became much weaker while the Au L lines became more prominent.

Fig. 8

(a) Logarithmically scaled image of the Perseus cluster. Its core is just on the aim point of the SXI. Two calibration areas are seen as bright regions on top and bottom center. The size of the image is $38\times 38\mathrm{arc}{\min }^2$.⁶³ (b) A logarithmically scaled image of the Crab Pulsar for 10- to 50-keV energy band, extracted from the HXI1 data of the Crab observation. The square in the image corresponds to the $9.1\times 9.1\mathrm{arc}{\min }^2$ FOV of the HXI. Note that the correction of the satellite attitude is preliminary and vignetting correction is not applied.⁶⁷

4.5.

Hard X-ray Imager

There are two HXI sensor modules: HXI1-S and HXI2-S [Fig. 7(b)]. The imager part of the HXI^67–74 consisted of four layers of 0.5-mm-thick double-sided silicon strip detectors (DSSD) and one layer of newly developed 0.75-mm-thick CdTe double-sided cross-strip detector (CdTe-DSD). In this configuration, soft x-ray photons below $\sim 30\mathrm{keV}$ are absorbed in the Si part, and hard x-ray photons above $\sim 30\mathrm{keV}$ go through the Si part and are detected by the CdTe part. To reduce the background, the sensor part is surrounded by a thick active shield and collimator made of bismuth germanate (BGO) scintillator coupled to avalanche photodiodes (APDs). In addition to the increase in efficiency, the stack configuration and individual readouts per each layer provide information on the interaction depth. This depth information is very useful for reducing the background in space applications, because we can expect that low-energy x-rays interact in the upper layers and, therefore, it is possible to reject the low-energy events, mainly due to the NXB, detected in lower layers. The $E<30\mathrm{keV}$ spectrum, obtained with the Si DSSD, has a much lower background due to the absence of activation in heavy material, such as Cd and Te. The DSSDs cover the energy below 30 keV while the CdTe strip detector covers the 30- to 80-keV band.

The electronics boxes of the HXIs (HXI-DE, DPU, and AE) were powered on February 28 to 29. Then, the HXI sensor (HXI-S) parts temperature was gradually decreased from 5°C to $-25°\mathrm{C}$ followed by the bias voltage operation of HXI1-S starting on March 8. The APD bias, Si DSSD bias, and CdTe DSD bias were raised to their operational voltage step by step, and the HXI1-S became operational on March 12. The HXI2-S followed on March 14. All 1280 readout channels for each of the HXI1-S and HXI2-S imager performed well, showing noise performance consistent with the prelaunch on-ground measurements. An image of the Crab Pulsar taken with HXI1 is shown in Fig. 8(b).^67,75

In addition to the thick BGO active shield and the concept of multilayer configuration of the imager, CXB baffling within the detector and inside the spacecraft worked well to reduce the background. The HXI achieved a low background level of $1-3\times {10}^{-4}\mathrm{cnt}{\mathrm{s}}^{-1}{\mathrm{cm}}^{-2}$ throughout their energy band of 5 to 80 keV, except for at a few energy region of activation lines in CdTe (here, the flux is normalized by the geometrical size of the detector, $10{\mathrm{cm}}^2$), which is the lowest level ever achieved in orbit. Due to the optimum design of the HXT and the long focal length of 12 m, the effective area was also the largest among the hard x-ray imaging spectroscopy instruments, flown to date. The HXI showed the potential to provide the highest sensitivity in this energy band, especially for diffuse sources.

4.6.

Soft Gamma-Ray Detector

Figure 7(c) shows a photograph of the SGD detector module. The SGD^76–81 measures soft $\gamma$-rays via reconstruction of Compton scattering in the semiconductor Compton camera, covering an energy range of 60 to 600 keV with a sensitivity at 300 keV, being 10 times better than that of the Suzaku Hard X-ray Detector,^82,83 by adopting a new concept of a narrow-FOV Compton telescope, combining Compton cameras and active well-type shields. There are two SGD sensor modules: SGD1 and SGD2. The Compton camera is made of 32 layers of 0.625-mm-thick Si pad detectors and 8 layers of 0.75-mm-thick CdTe pad detectors. The side of Si detectors is also surrounded by two layers of CdTe detectors. In the Si/CdTe Compton camera, events involving the incident $\gamma$-ray being scattered in the Si detectors and fully absorbed in the CdTe detectors are used for Compton imaging. The direction of the $\gamma$-ray is calculated by solving the Compton kinematics with information concerning deposit energies and interaction positions recorded in the detectors. In principle, each layer could act not only as a scattering part but also as an absorber part. A very compact, high-angular resolution (fineness of image) camera is realized by fabricating semiconductor imaging elements made of Si and CdTe, which have excellent performance in position resolution, high-energy resolution, and high-temporal resolution.

The camera was then mounted inside the bottom of a well-type active shield, made of BGO scintillators. The major advantage of employing a narrow FOV is that the direction of incident $\gamma$-rays is constrained to be inside the FOV. If the Compton cone, which corresponds to the direction of incident gamma rays, does not intercept the FOV, we can reject the event as background. Most of the background can be rejected by requiring this condition. The opening angle provided by the BGO shield is $\sim 10\mathrm{deg}$ at 500 keV. An additional PCuSn collimator restricts the field of view of the telescope to 33′ for photons below 100 keV, to minimize the flux due to the cosmic x-ray background in the FOV.

The power-on operation of SGD-DE, SGD-DPU, and SGD-AE was performed on March 1, 2016.^81,84 After this operation, SGD HK telemetries, including temperature monitors of SGD-S, were generated. Then, the SGD-S was gradually cooled down to $-25°\mathrm{C}$ from March 3 to 13 by changing the SGD heater settings. Start-up operation of SGD-S began on March 15, 2016. First, Compton cameras, APD CSAs for BGO readout, and HV modules for Compton cameras and APDs were powered on, and then we applied the high-voltage step by step and set up detectors into the nominal observation mode. From March 21, SGD1-S had been operated in the nominal observation mode, and SGD2-S was shifted into the nominal observation on March 24, 2016. Basic performance of the SGD was confirmed in the Crab observation, albeit with very short exposure.