PolarLight: a CubeSat X-ray Polarimeter based on the Gas Pixel Detector
Hua Feng, Weichun Jiang, Massimo Minuti, Qiong Wu, Aera Jung, Dongxin Yang, Saverio Citraro, Hikmat Nasimi, Jiandong Yu, Ge Jin, Jiahui Huang, Ming Zeng, Peng An, Luca Baldini, Ronaldo Bellazzini, Alessandro Brez, Luca Latronico, Carmelo Sgrò, Gloria Spandre, Michele Pinchera, Fabio Muleri, Paolo Soffitta, Enrico Costa
EExp Astron manuscript No. (will be inserted by the editor)
PolarLight: a CubeSat X-ray Polarimeter based on the GasPixel Detector
Hua Feng · Weichun Jiang · Massimo Minuti · Qiong Wu · AeraJung · Dongxin Yang · Saverio Citraro · Hikmat Nasimi · JiandongYu · Ge Jin · Jiahui Huang · Ming Zeng Peng An · Luca Baldini · Ronaldo Bellazzini · Alessandro Brez · Luca Latronico · CarmeloSgr`o · Gloria Spandre · Michele Pinchera · Fabio Muleri · PaoloSoffitta · Enrico Costa Received: date / Accepted: date
Abstract
The gas pixel detector (GPD) is designedand developed for high-sensitivity astronomical X-raypolarimetry, which is a new window about to open ina few years. Due to the small mass, low power, andcompact geometry of the GPD, we propose a Cube-Sat mission Polarimeter Light (PolarLight) to demon-strate and test the technology directly in space. Thereis no optics but a collimator to constrain the field ofview to 2.3 degrees. Filled with pure dimethyl ether(DME) at 0.8 atm and sealed by a beryllium windowof 100 µ m thick, with a sensitive area of about 1.4 mmby 1.4 mm, PolarLight allows us to observe the bright-est X-ray sources on the sky, with a count rate of, e.g., ∼ − from the Crab nebula. The PolarLightis 1U in size and mounted in a 6U CubeSat, whichwas launched into a low earth Sun-synchronous orbiton October 29, 2018, and is currently under test. Morelaunches with improved designs are planned in 2019.These tests will help increase the technology readinessfor future missions such as the enhanced X-ray Timingand Polarimetry (eXTP), better understand the orbitalbackground, and may help constrain the physics withobservations of the brightest objects. H. FengE-mail: [email protected] Department of Engineering Physics and Center for Astro-physics, Tsinghua University, Beijing 100084, China · KeyLaboratory for Particle Astrophysics, Institute of High En-ergy Physics, Chinese Academy of Sciences, Beijing 100049 · INFN-Pisa, Largo B. Pontecorvo 3, 56127 Pisa, Italy · School of Electronic and Information Engineering, NingboUniversity of Technology, Ningbo, Zhejiang 315211, China · North Night Vision Technology Co., Ltd., Nanjing 211106,China · IAPS/INAF, Via Fosso del Cavaliere 100, 00133Rome, Italy
Keywords astronomy · X-ray polarimetry · gas pixeldetector · CubeSat
X-ray polarimetry in the keV band has drawn greatinterests in astrophysics [9] but has remained unex-plored since 1970s [20,22,24]. Along with the break-through in detection technology that makes 2D electrontracking possible [7,5,3,2], high-sensitivity X-ray po-larimetry becomes possible and a number of space mis-sions dedicated to or capable of X-ray polarimetry havebeen proposed [15,8,23,25]. Active missions or missionconcepts include the Imaging X-ray Polarimetry Ex-plorer (IXPE) [23], which was selected by NASA andis scheduled to launch around 2021, and the enhancedX-ray Timing and Polarimetry (eXTP) [25], which isa Chinese-European collaboration aiming for large-areaX-ray polarimetry jointly with timing and spectroscopy.Both IXPE and eXTP have adopted the gas pixel de-tector (GPD) as the focal plane detector. Thus, a flighttest of the detector is indeed needed.The GPD polarimeter has the advantage of com-pactness and low mass, and can be operated at room-temperature with a total power of about 2 W. Theseindicate that it can easily fit into a CubeSat space-craft. Thus, based on the GPD tested in the lab [5,11],we modified the design and interface to be compati-ble with a 6U CubeSat developed by Spacety Co. Ltd,trying to test the technique in orbit (see Figure 1). Thedetector occupies a standard unit of the CubeSat and isnamed Polarimeter Light (PolarLight), but it is not theonly payload of the CubeSat. The satellite was success-fully launched into a nearly circular Sun-synchronousorbit on October 29, 2018, with an altitude of about a r X i v : . [ a s t r o - ph . I M ] M a r Feng et al.
520 km and an orbital period of 95 minutes. The Cube-Sat is equipped with a star tracker and three reactionwheels, enabling a pointing accuracy of about 0.1 ◦ to acelestial object. On November 16, the PolarLight waspowered on briefly and tested with charge injection (in-jection of charge to the preamplifier of a pixel), indicat-ing that the electronic part is working. On December18, the high voltage was applied for the first time andtracks triggered by X-rays and charged particles wereseen. After roughly three months since launch, tests forthe data transfer and attitude control were successfullydone with the CubeSat. At the time of writing, the Po-larLight is able to point at a celestial target and startobservations. Here, we describe the structural and elec-tronic design of PolarLight and the ground calibrationresults. The in-orbit tests will be reported in follow-uppapers. PolarLight contains three printed circuit boards (PCBs)inside an aluminum case. From top to bottom, the threePCBs are respectively to host the GPD, the high volt-age (HV) power supply and dividing circuits, and thedata acquisition (DAQ) system. The structure of thePolarLight is shown in Figure 2. Here, we elaborate thedetailed specifications of some key components.2.1 DetectorThe GPD for PolarLight is based on the design intro-duced by the INFN-Pisa group [5], and is similar tothe one reported in Li et al. [11]. It is a 2D gas pro-portional counter with pixel readout to measure thetrack image of photoelectrons emitted following the ab-sorption of X-rays. A valid event consists of followingprocesses. An incident X-ray goes though one of theholes of the collimator, penetrate the beryllium win-dow, and is absorbed by the working gas of the GPD.After absorption, a photoelectron is emitted and startsto ionize the gas molecules into ions and primary elec-trons. The primary electrons drift toward the anodeunder a paralleled electric field. When the electrons gothrough the holes of the gas electron multiplier (GEM),where the electric field is strong, avalanche happens andsecondary electrons are created. The electrons are en-hanced in number by a factor of a few hundred to a fewthousand (the gain factor). Then, the secondary elec-trons will drift along the field lines. Some end at thebottom electrode of the GEM, while the others can goall the way to the anode, which is the ASIC chip. Theinduced charge on the pixels will trigger the electronics + X + Y + Z + Y + X + Z Fig. 1: Schematic drawing of the CubeSat. The Po-larLight is mounted at the side away from the solarpanel, with the window pointing toward the + Y direc-tion.and be integrated, amplified, and filtered by the front-end electronics inside the chip. This is the whole chainhow an event is detected.Import components of the GPD include the follows. • Collimator — The collimator is a capillary platemade of lead glass (with ∼
38% lead oxide), manu-factured by North Night Vision Technology Co. Ltd.It is 1.66 mm thick and contains cylindrical microp-ores with a diameter of 83 µ m and an open fractionof 71%. The field of view (FOV) has a full width athalf maximum (FWHM) of 2 . ◦ and a full width atzero response (FWZR) of 5 . ◦ . A microscopic viewof the collimator is displayed in Figure 3. Using anX-ray tube (Oxford Instruments Apogee 5500 serieswith a focal spot size of 35 µ m) placed at ∼ • Window — The GPD is sealed by a beryllium win-dow of 100 µ m thick. The window is glued and elec-trically contacted to the titanium frame. When theGPD is in operation, the beryllium window and tita-nium frame are supplied with a HV of about − µ m thick coated withaluminum and will further absorb soft X-rays. TheAluminum coating has an unknown thickness, buttests with X-ray sources suggest that the absorptionis dominated by the mylar foil. olarLight 3 GPD boardHV boardDAQ board collimatoron an Al frameBe windowon a Ti frameceramic spacerGEMASIC
Fig. 2: Structure of the PolarLight and the GPD. ( deg r ee ) Fig. 3:
Top : microscopic image of the collimator. Thecollimator is 1.66 mm thick. The cylindrical apertureshave a diameter of 83 µ m and an open fraction of 71%.The FOV has a FWHM of 2 . ◦ and a FWZR of 5 . ◦ . Bottom : measured angular response of the collimatorusing an X-ray beam. The solid line is the data contourat half of the maximum response. The dotted lines rep-resents the designed FWHM and FWZR, respectively. Fig. 4: Microscopic image of the GEM foil, which is 100 µ m thick. The holes have a diameter of 50 µ m and apitch of 100 µ m.Fig. 5: Microscopic image of the ASIC chip around thecorner. The hexagonal pixels have a pitch of 50 µ m. Feng et al. driftmonitor HV HV Fig. 6: High voltage circuits diagram. There are two HVmodules for a cold backup. • GEM — The GEM offers signal amplification bymultiplying the number of primary electrons by afactor of a few hundred to a few thousand when theypass through the GEM holes. The GEM foil is a 100 µ m thick liquid crystal polymer (LCP) coated with5 µ m copper on both sides [19], manufactured bySciEnergy Inc., Japan, see Figure 4. The foil consistsof laser etched micro-holes with a diameter of 50 µ mand a pitch of 100 µ m in a hexagonal pattern. Theoperating high voltage ranges from 600–700 V acrossthe top and bottom electrodes, which determinesthe effective gain in an exponential law. • ASIC — The ASIC chip is used for collecting andprocessing the multiplied charge signals. Figure 5shows the top metal layer of the chip, which is pix-elated to hexagonal pixels. Each pixel is connectedto a full electronic chain (preamplifier, shaping am-plifier, sample and hold, and multiplexer) built im-mediately below it. The noise is around 50 e − rmsper pixel. The shaping time is 3–10 µ s and exter-nally adjustable. More details about the ASIC canbe found in Bellazzini et al. [4]. • Gas — The detection gas sealed in the GPD is pure( > Component Mass (g)GPD board 121HV board 60DAQ board 59Aluminum case 322Accessaries (screws/wires/etc.) 19Total 581 geometry with a length of 0.5 inch on each side, and ispowered by a low voltage power supply of 5 V. With aprogramming pin, the HV output is adjustable from 0to − ∼
580 g and thetotal power consumption is about ∼ olarLight 5 GPD board HV boardDAQ boardBe windowGEM top
Analog outputReference voltageDigital I/O config & readout
Temperaturesensor BADCDAC
HV moduleDividing circuitsASICconfiguration
GPDASIC
Temperaturesensor A
MCU
Science event managementHouse keeping management Comm. & data transferFlash memory (64 Mbytes)Internal RAM(512 kbytes)HVmanagementLV currents monitor
GEM bottom SPI
Fig. 7: Diagram for the back-end electronics of PolarLight.Table 2: Power consumptions of PolarLight
HV Trigger 5V power (W) 3.3V power (W) Total power (W)off off 1.45 0.78 2.230 V off 1.48 0.78 2.260 V charge injection 1.46 0.52 1.98 ∼ ∼ − in the detector. The typical number ofpixels for events triggered by 4 keV X-rays is about 700.Thus, the total data rate is at least 20 Mbytes per hourif we point the detector at Scorpius X-1. The Cube-Sat is operated by Spacety. The commands, telemetry,and small data packages can be transferred through theUHF channel. There are two UHF ground stations forSpacety with 4–6 times of overflight every day. The fulldata will be transferred to the ground station via theX band, with a chance expected roughly once a week.2.6 Space qualification testsWe conducted most of the qualification tests for spaceenvironment before the payload was delivered to thesatellite company for integration, including the mechan-ical and thermal tests. Due to a tight schedule, the Feng et al.
Table 3: Science data structure of PolarLight.
BytesHeader 4Time 24ROI 8Image n pixel × n pixel × D e t e c t i on e ff i c i en cy with thermal coatwithout thermal coat Fig. 8: Detection efficiency of PolarLight. The dashedline represents the efficiency for the GPD, while thesolid line is the efficiency after the thermal coat is cov-ered.thermal-vacuum test was not done alone, but alongwith the whole CubeSat after integration. In the thermal-vacuum chamber, an Fe source was used to monitorthe detector performance and the results are as ex-pected. Right after each mechanical test, a resonancesearch was conducted to detect whether or not therewas a frequency shift due to mechanical deformation.The test conditions are summarized in Table 4.
The detection efficiency of the detector is determinedby the thermal coat (6 µ m mylar), the beryllium win-dow (100 µ m), and the working gas (1 cm thick DMEat a pressure of 0.8 atm). The thermal coat is optional,but we decided to put it on in order for a stable temper-ature control. A calculation of the detection efficiencyis shown in Figure 8.The detector was filled with the working gas andsealed on August 20th, 2018. The gain was found to in-crease rapidly with time. This is due to the fact that thechamber is sufficiently pumped and the materials inside P ea k P o s i t i on ( P H A ) G r ound w i r e c hanged C oppe r t ube c u t in the labthermal vacuumintegrated Fig. 9: Time variation of the detector gain since the sealof the GPD on August 20th, 2018. The dashed linesindicate the times when the ground wire was changedand the copper tube was cut, respectively. The payloadwas first tested alone in the lab, and then integratedinto the CubeSat.are adequately degassed, such that our working gas willbe absorbed by the materials, especially by those witha relatively high outgassing rate, until an equilibrium isapproached. The gain variation curve is shown in Fig-ure 9, where we have converted the peak position tothat of Fe with a HV of 3200 V if the measurementwas not done in that case. The electric interface be-tween the payload to the CubeSat is a 21-pin connector,for both power and communication. About 6 days af-ter the detector seal, we realized that a single wire forthe ground was insufficient to damp the power surgecaused by the HV module, and the HV ground waspulled higher when the HV was on. Thus, we added asecond ground wire to solve the problem. This also leadto an increase of the gain as the HV ground was stabi-lized at zero. About 7 days after the seal, the detectorworked stably and we cut the copper tube, which wasused to pump the chamber and fill the gas, to its min-imum size. That action may have compressed the gasin the chamber so that the gain dropped (the gain isinversely scaled with the pressure in our case). Then,laboratory tests and calibrations were conducted be-fore the payload was shipped to the satellite companyfor integration around 15 days after seal. Since then,we were not able to test the payload with X-rays for awhile except on the 30th day when the whole CubeSatwas tested for thermal-vacuum qualification. About oneweek before the CubeSat was shipped to the launch site,we were allowed to test it and measured a few spectrato verify the detector status. The test was done with olarLight 7
Table 4: Qualification tests for space environment.
Test Date ConditionsRandom 2018 Aug 05 10–2000 Hz, 8.2 g (rms), 2 minsSinusoidal 2018 Aug 05 0–100 Hz, 1.5 g, 2.5 minsShock 2018 Aug 09 1000 g, twice in each directionThermal 2017 July 24-29 −
15 to +45 ◦ C, 12.5 cyclesThermal-vacuum 2018 Sep 18-22 − ◦ C, 3.5 cycles, < − Pa Y ( mm ) C oun t s before correctionafter correction Fig. 10:
Top : gain map (70 ×
70) of the GPD acrossthe sensitive surface, normalized to the mean peak po-sition measured with the Fe source. A value largerthan unity means the gain in that area is higher thanthe mean, and vice versa. Pixels around the edge havea low gain because of the loss of charges.
Bottom : Fespectra before and after correction with the gain map.The FWHM to mean ratio changes from 33.5% to 18.6%after the gain map correction.a laboratory power supply at first. Then, the batteryfrom the satellite was used (the last point in Figure 9).As one can see, the gain increased significantly. Thisis probably because the low voltages provided by thesatellite battery are not as accurate as expected, andthe output PHA is affected. The phenomenon can berepeated using a backup system in the lab. This should be improved in the future by adding a DC-DC modulein the payload. Fortunately, the gain variation is lessthan the spectral resolution ( ∼ ≤ Fe source, we found that thegain is not uniform across the detector surface. This isdue to the non-uniformity of the GEM thickness [18].We divided the sensitive area into 70 ×
70 cells, eachwith a size of 0 . × . Fe source and measured thepeak position in each cell to reflect the gain variation.From here on, we select events that have only one clus-ter and the cluster size is at least 45 pixels for analysis.A gain map, normalized to the mean peak position ofall cells, is created and shown in Figure 10. If the gainmap is applied on the Fe spectrum, the FWHM tomean ratio changes from 33.5% to 18.5%, suggestive ofa successful correction. Then, the level-2 file adds thegain map corrected PHA for each event.We use four Bragg crystals and their 45-degree diffrac-tions to measure the energy spectra and modulationfactors of the detector. A silicon PIN detector was usedto take the diffracted energy spectra to check whetheror not the peaks appear at the energies as expected.The measured energy spectra with the PolarLight areshown in Figure 11. The diffraction energies, along withthe measured energy resolutions are listed in Table 5.The fractional energy resolution (FWHM/ E ) follows an Feng et al.
Table 5: Bragg crystals, diffracted energies at 45 de-grees, and the measured energy resolutions and modu-lation factors.
Crystal Order E FWHM/
E µ (keV)PET I 2.01 0 . ± .
008 0 . ± . I 2.67 0 . ± .
002 0 . ± . . ± .
001 0 . ± . II 5.33 0 . ± .
002 0 . ± . . ± .
001 0 . ± . . ± .
006 0 . ± . PE T - I ( . k e V ) M g F - I ( . k e V ) Al-I (3.74 keV) M g F - II ( . k e V ) L i F - II ( . k e V ) A l - II ( . k e V ) Fig. 11: Energy spectra measured with 45-degree Braggdiffractions. E − / relation except at the lowest energy, where theenergy resolution is smaller than expected, but this maybe due to the low statistics and large error associatedwith the PET measurement.As 45-degree diffraction produces a fully polarizedX-ray beam, the same data are used to calculate themodulation factors. Following the literature [14,11], wediscarded 25% of the events with the lowest eccentric-ity. The emission angle distributions of electrons areshown in Figure 12, and fitted with a modulation func-tion, A + B cos ( φ − φ ). At energies below 3 keV, themajor axis direction is adopted as the emission angle,while above 3 keV, the impact point method is used.The degree of modulation, (max − min) / (max + min),is then calculated and displayed in Table 5 and Fig-ure 13 (top panel), which is the modulation factor ( µ ;degree of modulation resulted from fully polarized X-rays) of the instrument. As the diffracted beam cannotilluminate the whole detector plane at a close distance,we thus checked the positional uniformity at 9 (3 × C oun t s PET-I (2.01 keV)90 60 30 0 30 60 90Angle (degree)2.50.02.5 150200250300 C oun t s MgF2-I (2.67 keV)90 60 30 0 30 60 90Angle (degree)2.50.02.5200300400500600700 C oun t s Al-I (3.74 keV)90 60 30 0 30 60 90Angle (degree)2.50.02.5 100150200250300350400450 C oun t s MgF2-II (5.33 keV)90 60 30 0 30 60 90Angle (degree)2.50.02.5100150200250300350400450 C oun t s LiF-II (6.14 keV)90 60 30 0 30 60 90Angle (degree)2.50.02.5 50100150200 C oun t s Al-II (7.49 keV)90 60 30 0 30 60 90Angle (degree)2.50.02.5
Fig. 12: Modulation curves measured with 45-degreeBragg diffractions.the modulation factor and position angle do not showa detectable change with respect to the location.The sensitivity of an X-ray polarimeter is propor-tional to the quality factor, which is a product of themodulation factor and the square root of the detectionefficiency. In Figure 13 (bottom panel), we display thequality factor as a function of energy. As one can see,the sensitivity of the PolarLight peaks around 4 keV.We note that our results are well consistent withthose reported by previous studies, except at 2.7 keVfor the MgF -I line, where a modulation factor of 0.27was reported [14,13,11]. The major difference betweenour setup and the previous is the GEM pitch. For Po-larLight, the GEM pitch is 100 µ m and larger thanbefore. For X-rays of higher energies, the electron trackis long and the result may not be limited by the pitch,while for the lowest energy (2 keV), the track is unre-solved so that the pitch is not important. Around 3 keV,the pitch may have played the most important rolein sampling the electron track. This may explain whyconsistent modulation factors can be obtained at otherenergies. More experiments and simulations should bedone in the future to investigate this problem. olarLight 9 M odu l a t i on f a c t o r Q ua li t y f a c t o r Fig. 13: Modulation factor ( top ) and the quality factor( bottom ) for PolarLight. The quality factor is definedas modulation factor times the square root of the effi-ciency.
As the boarder region may suffer from charge loss (seethe gain map in Figure 10), we only extract the ± α line originatedfrom the GEM foil appears above this band. The 3–5keV band is the energy range where the sensitivity is thehighest, and the 4–8 keV band has the highest modula-tion factor. The average modulation factor is 0.25, 0.37,and 0.49, respectively in the three bands, weighted us-ing a detected Crab spectrum. The choice of the energyband relies on the specific scientific objective.The background rate due to the CXB is found tobe a few times 10 − counts s − and is negligible com-pared with the expected flux from the brightest X-raysources. However, as the CubeSat is in a polar orbit,where the particle flux is high near the polar and the
180 120 60 0 60 120 180Longitude (degree)9060300306090 La t i t ude ( deg r ee ) t r apped e l e c t r on f l u x ( E > k e V ; c m s ) = 4 . µS (cid:114) S + BT , (1)where S is the source count rate, B is the backgroundcount rate, µ the modulation factor, and T is the to-tal exposure time. For the Crab nebula, which has aknown degree of polarization of 19%, its polarizationsignal in the 2–8 keV band can be detected with an ex-posure time of a few times 10 s if the background rate is ∼ − . If the background exceeds 5 counts s − ,the polarization from Crab is no longer detectable evenwith a net exposure of 10 s. Because the HV cannotbe powered on in regions of high particle flux, also dueto Earth occultation, the effective observing time is lessthan one half of the total operation time.Anyway, bright X-ray sources on the sky, includ-ing pulsar wind nebulae and accreting compact objects,could be targets of PolarLight (Figure 15). We note thatstrong magnetic systems, such as accreting pulsars, areof particular interest because of their potentially highdegree of polarization. Limited by the power, some re-gions (shaded in Figure 15) in the sky may be pointedby the PolarLight continuously. While other regions,due to large angles between the solar panel and the Table 6: Expected count rates for bright X-ray sources in different energy bands measured with PolarLight.
Rate (counts s − ) Reference(2–8 keV) (3–5 keV) (4–8 keV)Crab 0.20 0.068 0.029 [10]GRS 1915+105 (thermal state) 0.16 0.078 0.040 [12]Cygnus X-1 (low/hard state) 0.078 0.029 0.014 [17]Scorpius X-1 3.5 1.2 0.46 [6] D e c li na t i on ( deg r ee ) CrabCyg X-1GRS 1915+105 Sco X-1GX 5-1GX 17+2GX 9+1 Her X-1 Vela X-1Cen X-3
Power optimal BHB PWN NS LMXB Accreting pulsar
Fig. 15: Possible targets for PolarLight, including blackhole binaries (BHBs), pulsar wind nebulae (PWNe),neutron star low-mass X-ray binaries (NS LMXBs), andaccreting pulsars (highly magnetized systems). If thesources are in the shaded region, they may be pointedby PolarLight continuously without a power issue. Forsources outside the shaded region, they can be observedbut the observation has to be interrupted due to bat-tery charge and/or Earth occultation due to a largeangle between the Sun and solar panel.Sun, the observation can be conducted but has to beintermittent due to battery charge and/or Earth occul-tation.
Here, we report on the design and ground test resultsfor PolarLight, which did the first flight test for theGPD polarimeter. The main purpose for PolarLight isto demonstrate the technique in space and reveal poten-tial issues with the detector design, which will be valu-able for future missions like eXTP as the same detectorwill be used. Limited by the tight schedule and con-straints on resources, some issues already emerged dur-ing the ground test and calibration. We list the lessonsgained so far: • A DC-DC module is necessary to provide a stablepower supply for the detector. • The detector needs be sealed at least a few monthsin advance, so that the gain will be stable at thetime of launch. • The gain is sensitive to the gas pressure. The equi-librium between the absorption and outgassing in-side the chamber is a function of temperature. Changeof the storage temperature may lead to a change ofthe gain temporarily. Once launched, the detectortemperature would best be controlled in a narrowrange, no matter in operation or not. • A gain calibration seems useful every time when anobservation is done. Online calibration is impossiblefor a CubeSat, but has been designed for future largemissions. • Whether or not the 100- µ m-pitch GEM can pro-duce a modulation factor as high as that with a50- µ m-pitch GEM needs in-depth investigations. Acoarse pitch allows for a thicker GEM (100 µ m) anda larger gain.After the CubeSat finishes the communication andattitude test, we will start the full test of the PolarLightand investigate the in-orbit background and its influ-ence to the polarization measurement. A new flight isplanned in 2019 with an improved design. A 50 µ mwindow will be used and known issues will be fixed. Acknowledgements
We thank Wenfei Yu for helpful dis-cussions about the choice of targets, and Spacety for help-ing with the test. HF acknowledges funding support fromthe National Natural Science Foundation of China under thegrant Nos. 11633003 and 11821303, and the National KeyR&D Program of China (grant Nos. 2018YFA0404502 and2016YFA040080X). The initial development of the GPD con-cept was funded by the Italian National Institute for NuclearPhysics (INFN), the Italian Space Agency (ASI), and theItalian National Institute for Astrophysics (INAF).
References
1. Bellazzini, R., Angelini, F., Baldini, L., Brez, A., Costa,E., Di Persio, G., Latronico, L., Massai, M.M., Omodei,N., Pacciani, L., Soffitta, P., Spandre, G.: Novel gaseusX-ray polarimeter: data analysis and simulation. In:olarLight 11S. Fineschi (ed.) Polarimetry in Astronomy,
Proc. SPIE , vol. 4843, pp. 383–393 (2003). DOI 10.1117/12.4593812. Bellazzini, R., Brez, A., Costa, E., Minuti, M., Muleri, F.,Pinchera, M., Rubini, A., Soffitta, P., Spandre, G.: Pho-toelectric X-ray Polarimetry with Gas Pixel Detectors.Nuclear Instruments and Methods in Physics ResearchA , 173–177 (2013). DOI 10.1016/j.nima.2012.12.0063. Bellazzini, R., Muleri, F.: X-ray polarimetry: A new win-dow on the high energy sky. Nuclear Instruments andMethods in Physics Research A , 766–770 (2010).DOI 10.1016/j.nima.2010.04.0064. Bellazzini, R., Spandre, G., Minuti, M., Baldini, L., Brez,A., Cavalca, F., Latronico, L., Omodei, N., Massai, M.M.,Sgro’, C., Costa, E., Soffitta, P., Krummenacher, F., deOliveira, R.: Direct reading of charge multipliers with aself-triggering CMOS analog chip with 105 k pixels at 50 µ m pitch. Nuclear Instruments and Methods in PhysicsResearch A , 552–562 (2006). DOI 10.1016/j.nima.2006.07.0365. Bellazzini, R., Spandre, G., Minuti, M., Baldini, L., Brez,A., Latronico, L., Omodei, N., Razzano, M., Massai,M.M., Pesce-Rollins, M., Sgr´o, C., Costa, E., Soffitta, P.,Sipila, H., Lempinen, E.: A sealed Gas Pixel Detectorfor X-ray astronomy. Nuclear Instruments and Meth-ods in Physics Research A , 853–858 (2007). DOI10.1016/j.nima.2007.05.3046. Church, M.J., Gibiec, A., Ba(cid:32)luci´nska-Church, M., Jack-son, N.K.: Spectral investigations of the nature of theScorpius X-1 like sources. Astron. Astrophys. , A35(2012). DOI 10.1051/0004-6361/2012189877. Costa, E., Soffitta, P., Bellazzini, R., Brez, A., Lumb, N.,Spandre, G.: An efficient photoelectric X-ray polarimeterfor the study of black holes and neutron stars. Nature , 662–665 (2001)8. Jahoda, K.M., Black, J.K., Hill, J.E., Kallman, T.R.,Kaaret, P.E., Markwardt, C.B., Okajima, T., Petre, R.,Soong, Y., Strohmayer, T.E., Tamagawa, T., Tawara, Y.:X-ray polarization capabilities of a small explorer mis-sion. In: Space Telescopes and Instrumentation 2014:Ultraviolet to Gamma Ray, Proc. SPIE , vol. 9144, p.91440N (2014). DOI 10.1117/12.20567199. Kallman, T.: Astrophysical motivation for X-ray po-larimetry. Advances in Space Research , 2673–2677(2004). DOI 10.1016/j.asr.2003.03.05910. Kirsch, M.G., Briel, U.G., Burrows, D., Campana, S.,Cusumano, G., Ebisawa, K., Freyberg, M.J., Guainazzi,M., Haberl, F., Jahoda, K., Kaastra, J., Kretschmar, P.,Larsson, S., Lubi´nski, P., Mori, K., Plucinsky, P., Pollock,A.M., Rothschild, R., Sembay, S., Wilms, J., Yamamoto,M.: Crab: the standard x-ray candle with all (modern)x-ray satellites. In: O.H.W. Siegmund (ed.) UV, X-Ray,and Gamma-Ray Space Instrumentation for AstronomyXIV, Proc. SPIE , vol. 5898, pp. 22–33 (2005). DOI10.1117/12.61689311. Li, H., Feng, H., Muleri, F., Bellazzini, R., Minuti, M.,Soffitta, P., Brez, A., Spandre, G., Pinchera, M., Sgr´o,C., Baldini, L., She, R., Costa, E.: Assembly and testof the gas pixel detector for X-ray polarimetry. NuclearInstruments and Methods in Physics Research A ,155–162 (2015). DOI 10.1016/j.nima.2015.09.06012. McClintock, J.E., Shafee, R., Narayan, R., Remillard,R.A., Davis, S.W., Li, L.X.: The Spin of the Near-Extreme Kerr Black Hole GRS 1915+105. Astrophys.J. , 518–539 (2006). DOI 10.1086/50845713. Muleri, F., Bellazzini, R., Brez, A., Costa, E., Fabiani, S.,Minuti, M., Pinchera, M., Rubini, A., Soffitta, P., Span-dre, G.: A new design for the gas pixel detector. In: Space Telescopes and Instrumentation 2012: Ultravioletto Gamma Ray,
Proc. SPIE , vol. 8443, p. 84430L (2012).DOI 10.1117/12.92600614. Muleri, F., Soffitta, P., Baldini, L., Bellazzini, R., Brez,A., Costa, E., Fabiani, S., Krummenacher, F., Latron-ico, L., Lazzarotto, F., Minuti, M., Pinchera, M., Ru-bini, A., Sgr´o, C., Spandre, G.: Spectral and polari-metric characterization of the Gas Pixel Detector filledwith dimethyl ether. Nuclear Instruments and Meth-ods in Physics Research A , 285–293 (2010). DOI10.1016/j.nima.2010.03.00615. Soffitta, P., Barcons, X., Bellazzini, R., Braga, J., Costa,E., Fraser, G.W., Gburek, S., Huovelin, J., Matt, G.,Pearce, M., Poutanen, J., Reglero, V., Santangelo, A.,Sunyaev, R.A., Tagliaferri, G., Weisskopf, M., Aloisio,R., Amato, E., Attin´a, P., Axelsson, M., Baldini, L.,Basso, S., Bianchi, S., Blasi, P., Bregeon, J., Brez, A.,Bucciantini, N., Burderi, L., Burwitz, V., Casella, P.,Churazov, E., Civitani, M., Covino, S., Curado da Silva,R.M., Cusumano, G., Dadina, M., D’Amico, F., De Rosa,A., Di Cosimo, S., Di Persio, G., Di Salvo, T., Dovciak,M., Elsner, R., Eyles, C.J., Fabian, A.C., Fabiani, S.,Feng, H., Giarrusso, S., Goosmann, R.W., Grandi, P.,Grosso, N., Israel, G., Jackson, M., Kaaret, P., Karas,V., Kuss, M., Lai, D., La Rosa, G., Larsson, J., Lars-son, S., Latronico, L., Maggio, A., Maia, J., Marin, F.,Massai, M.M., Mineo, T., Minuti, M., Moretti, E., Mu-leri, F., O’Dell, S.L., Pareschi, G., Peres, G., Pesce, M.,Petrucci, P.O., Pinchera, M., Porquet, D., Ramsey, B.,Rea, N., Reale, F., Rodrigo, J.M., R´o˙za´nska, A., Ru-bini, A., Rudawy, P., Ryde, F., Salvati, M., de Santi-ago, V.A., Sazonov, S., Sgr´o, C., Silver, E., Spandre,G., Spiga, D., Stella, L., Tamagawa, T., Tamborra, F.,Tavecchio, F., Teixeira Dias, T., van Adelsberg, M., Wu,K., Zane, S.: XIPE: the X-ray imaging polarimetry ex-plorer. Experimental Astronomy , 523–567 (2013).DOI 10.1007/s10686-013-9344-316. Strohmayer, T.E., Kallman, T.R.: On the StatisticalAnalysis of X-Ray Polarization Measurements. Astro-phys. J. , 103 (2013). DOI 10.1088/0004-637X/773/2/10317. Sugimoto, J., Mihara, T., Kitamoto, S., Matsuoka, M.,Sugizaki, M., Negoro, H., Nakahira, S., Makishima, K.:MAXI observations of long-term variations of Cygnus X-1 in the low/hard and the high/soft states. PASJ , S17(2016). DOI 10.1093/pasj/psw00418. Takeuchi, Y., Kitaguchi, T., Hayato, A., Tamagawa,T., Iwakiri, W., Asami, F., Yoshikawa, A., Kaneko, K.,Enoto, T., Black, K., Hill, J.E., Jahoda, K.: Propertiesof the flight model gas electron multiplier for the GEMSmission. In: Space Telescopes and Instrumentation 2014:Ultraviolet to Gamma Ray, Proc. SPIE , vol. 9144, p.91444N (2014). DOI 10.1117/12.205715919. Tamagawa, T., Hayato, A., Asami, F., Abe, K., Iwamoto,S., Nakamura, S., Harayama, A., Iwahashi, T., Kon-ami, S., Hamagaki, H., Yamaguchi, Y.L., Tawara, H.,Makishima, K.: Development of thick-foil and fine-pitchGEMs with a laser etching technique. Nuclear Instru-ments and Methods in Physics Research A , 390–396(2009). DOI 10.1016/j.nima.2009.07.01420. Weisskopf, M.C., Cohen, G.G., Kestenbaum, H.L., Long,K.S., Novick, R., Wolff, R.S.: Measurement of the X-ray polarization of the Crab Nebula. Astrophys. J. ,L125–L128 (1976). DOI 10.1086/18224721. Weisskopf, M.C., Elsner, R.F., O’Dell, S.L.: On under-standing the figures of merit for detection and mea-surement of x-ray polarization. In: Space Telescopes2 Feng et al.and Instrumentation 2010: Ultraviolet to Gamma Ray,
Proc. SPIE , vol. 7732, p. 77320E (2010). DOI 10.1117/12.85735722. Weisskopf, M.C., Kestenbaum, H.L., Long, K.S., Novick,R., Silver, E.H.: An upper limit to the linear X-ray po-larization of Scorpius X-1. Astrophys. J. , L13–L16(1978). DOI 10.1086/18265523. Weisskopf, M.C., Ramsey, B., O’Dell, S., Tennant,A., Elsner, R., Soffitta, P., Bellazzini, R., Costa, E.,Kolodziejczak, J., Kaspi, V., Muleri, F., Marshall, H.,Matt, G., Romani, R.: The Imaging X-ray PolarimetryExplorer (IXPE). In: Space Telescopes and Instrumenta-tion 2016: Ultraviolet to Gamma Ray,
Proc. SPIE , vol.9905, p. 990517 (2016). DOI 10.1117/12.223524024. Weisskopf, M.C., Silver, E.H., Kestenbaum, H.L., Long,K.S., Novick, R.: A precision measurement of the X-raypolarization of the Crab Nebula without pulsar contam-ination. Astrophys. J. , L117–L121 (1978). DOI10.1086/18264825. Zhang, S.N., Feroci, M., Santangelo, A., Dong, Y.W.,Feng, H., Lu, F.J., Nandra, K., Wang, Z.S., Zhang, S.,Bozzo, E., Brandt, S., De Rosa, A., Gou, L.J., Hernanz,M., van der Klis, M., Li, X.D., Liu, Y., Orleanski, P.,Pareschi, G., Pohl, M., Poutanen, J., Qu, J.L., Schanne,S., Stella, L., Uttley, P., Watts, A., Xu, R.X., Yu, W.F.,in ’t Zand, J.J.M., Zane, S., Alvarez, L., Amati, L., Bal-dini, L., Bambi, C., Basso, S., Bhattacharyya S., R., B.,Belloni, T., Bellutti, P., Bianchi, S., Brez, A., Bursa, M.,Burwitz, V., Budtz-Jørgensen, C., Caiazzo, I., Campana,R., Cao, X., Casella, P., Chen, C.Y., Chen, L., Chen, T.,Chen, Y., Chen, Y., Chen, Y.P., Civitani, M., Coti Ze-lati, F., Cui, W., Cui, W.W., Dai, Z.G., Del Monte, E.,de Martino, D., Di Cosimo, S., Diebold, S., Dovciak, M.,Donnarumma, I., Doroshenko, V., Esposito, P., Evange-lista, Y., Favre, Y., Friedrich, P., Fuschino, F., Galvez,J.L., Gao, Z.L., Ge, M.Y., Gevin, O., Goetz, D., Han,D.W., Heyl, J., Horak, J., Hu, W., Huang, F., Huang,Q.S., Hudec, R., Huppenkothen, D., Israel, G.L., Ingram,A., Karas, V., Karelin, D., Jenke, P.A., Ji, L., Korpela,S., Kunneriath, D., Labanti, C., Li, G., Li, X., Li, Z.S.,Liang, E.W., Limousin, O., Lin, L., Ling, Z.X., Liu, H.B.,Liu, H.W., Liu, Z., Lu, B., Lund, N., Lai, D., Luo, B.,Luo, T., Ma, B., Mahmoodifar, S., Marisaldi, M., Martin-dale, A., Meidinger, N., Men, Y., Michalska, M., Mignani,R., Minuti, M., Motta, S., Muleri, F., Neilsen, J., Or-landini, M., Pan, A.T., Patruno, A., Perinati, E., Pic-ciotto, A., Piemonte, C., Pinchera, M., Rachevski A.,Rapisarda, M., Rea, N., Rossi, E.M.R., Rubini, A., Sala,G., Shu, X.W., Sgro, C., Shen, Z.X., Soffitta, P., Song,L., Spandre, G., Stratta, G., Strohmayer, T.E., Sun, L.,Svoboda, J., Tagliaferri, G., Tenzer, G., Hong, T., Tav-erna, R., Torok, G., Turolla, R., Vacchi, S., Wang, J.,Walton, D., Wang, K., Wang, J.F., Wang, R.J., Wang,Y.F., Weng, S.S., Wilms, J., Winter, B., Wu, X., Wu,X.F., Xiong, S.L., Xu, Y.P., Xue, Y.Q., Yan, Z., Yang, S.,Yang, X., Yang, Y.J., Yuan, F., Yuan, W.M., Yuan, Y.F.,Zampa, G., Zampa, N., Zdziarski, A., Zhang, C., Zhang,C.L., Zhang, L., Zhang, X., Zhang, Z., Zhang, W.D.,Zheng, S.J., Zhou, P., Zhou X. L.: eXTP: Enhanced X-ray Timing and Polarization mission. In: Space Tele-scopes and Instrumentation 2016: Ultraviolet to GammaRay,