Mitani et al. Earth, Planets and Space (2018) 70:77 https://doi.org/10.1186/s40623-018-0853-1

FULL PAPER Open Access High‑energy electron experiments (HEP) aboard the ERG (Arase) satellite Takefumi Mitani1*, Takeshi Takashima1, Satoshi Kasahara2, Wataru Miyake3 and Masafumi Hirahara4

Abstract This paper reports the design, calibration, and operation of high-energy electron experiments (HEP) aboard the exploration of energization and radiation in geospace (ERG) satellite. HEP detects 70 keV–2 MeV electrons and gener- ates a three-dimensional velocity distribution for these electrons in every period of the satellite’s rotation. Electrons are detected by two instruments, namely HEP-L and HEP-H, which difer in their geometric factor (G-factor) and range 4 2 of energies they detect. HEP-L detects 70 keV–1 MeV electrons and its G-factor is 9.3 10− ­cm sr at maximum, 3 2 × while HEP-H observes 0.7–2 MeV electrons and its G-factor is 9.3 10− ­cm sr at maximum. The instruments utilize silicon strip detectors and application-specifc integrated circuits× to readout the incident charge signal from each strip. Before the launch, we calibrated the detectors by measuring the energy spectra of all strips using γ-ray sources. To evaluate the overall performance of the HEP instruments, we measured the energy spectra and angular responses with electron beams. After HEP was frst put into operation, on February 2, 2017, it was demonstrated that the instru- ments performed normally. HEP began its exploratory observations with regard to energization and radiation in geospace in late March 2017. The initial results of the in-orbit observations are introduced briefy in this paper. Keywords: ERG, Arase, Radiation belts, High-energy electrons

Introduction Te exploration of energization and radiation in geo- Te mechanisms by which electrons are accelerated space (ERG) project explores the acceleration, transpor- in geospace are a key research area in solar-terrestrial tation, and loss of relativistic electrons in the radiation plasma physics. Relativistic electrons are trapped in belts and the dynamics of storms in geospace (Miyoshi the outer Van Allen radiation belt. Te electron fux is et al. 2012; Miyoshi et al. in review). To gain a detailed known to decrease rapidly during the main phase of the understanding of the acceleration and transport pro- magnetic storms, and then increases during the storms’ cesses, the electrons must be observed over a wide range recovery phase (Baker et al. 1986; Nagai 1988; Reeves of energies and electromagnetic felds with a wide range et al. 2003). Te processes causing fux variation must of frequencies. Te experiment observes a wide range be investigated with comprehensive in situ observations. of geospace particles, including the warm electrons, hot Particle acceleration occurs in various environments electrons of the plasma sheet, and sub-relativistic and around astronomical objects such as supernova remnants relativistic electrons of the radiation belts. Te satellite and pulsar-wind nebulae (Makishima 1999). Moreover, uses four instruments, namely LEP-e, MEP-e, HEP, and the Earth’s radiation belts ofer a unique opportunity to XEP, in order to measure electrons over this wide range observe these particles and waves in situ. Armed with of energies (Kazama et al. 2017; Kasahara et al. 2018a; an understanding of the acceleration mechanism in geo- Higashio et al. in review). Te high-energy electron space, some insight might be gained into the phenomena experiments (HEP) onboard the ERG satellite detects of electron acceleration in other astronomical objects. 70 keV–2 MeV electrons and generates a three-dimen- sional velocity distribution of electrons for every period *Correspondence: [email protected].jaxa.jp of the satellite’s spin. Tis energy range covers relativ- 1 Institute of Space and Astronautical Science, Japan Aerospace istic electrons and their seed electrons. Te full suite of Exploration Agency, Sagamihara 252‑5210, Kanagawa, Japan Full list of author information is available at the end of the article instruments aboard the ERG observes a wide range of

© The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat​iveco​mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Mitani et al. Earth, Planets and Space (2018) 70:77 Page 2 of 14

electrons. Tree additional devices have been installed: in the readout electronics, the above G-factor is a maxi- XEP (Higashio et al. in review) detects electrons with mum value. If we assume that the maximum electron 9 −2 energies of 0.4–20 MeV, MEP-e (Kasahara et al. 2018a) fux in the satellite’s orbit is 10­ × (E/[keV]) , then, detects electrons with energies of 7–87 keV, and LEP-e the total count rate is estimated to be 4.4 kHz above (Kazama et al. 2017) detects electrons with energies of 70 keV, and 440 Hz above 0.7 MeV by using the G-fac- 17 eV–20 keV. In MEP-e, the electrons are energy-fltered tor of the HEP-L module, for which the total count of by an electrostatic analyzer and the maximum detected electrons above 0.7 MeV in each eight-second rotation 3 energy is 87 keV. To allow for continuous energy cover- period is estimated to be 3.5 × 10 over 4π steradians. age, the lower detection range of HEP overlaps with the To detect more electrons at 0.7 MeV, the geometric fac- maximum MEP-e energy. tor of HEP-H is designed to be ten times larger than Te ERG satellite, which is also known as ‘Arase,’ was that of HEP-L. A photograph of the HEP fight model is launched from the Uchinoura Space Center at 11:00 on shown in Fig. 1. Tree HEP-L and three HEP-H modules December 20, 2016, by using the Epsilon launch vehicle. are housed in the hexagonal cylinder, and the electron- Te spacecraft attained an orbit with an apogee and peri- ics boards are housed in the black box. Each of the three gee altitude of approximately 32,000 and 400 km, respec- sides of the hexagonal cylinder has two slits, through tively. Tis allowed the satellite to record observations which the electrons enter the instrument. Figure 2a illus- over the entire extent of the radiation belts. After all the trates the HEP’s mounting onto the ERG satellite and instruments and functions were successfully checked, the marks its felds of view as gray areas corresponding to scientifc observations commenced in late March 2017. approximately 60° × 10°. Te HEP instrument is mounted Tis paper describes the design of the HEP instru- on the panel normal to the + X axis of the ERG satellite. ments, prelaunch testing, and initial results of the in- As the satellite rotates around the ­Zsc-axis, HEP covers orbit observations. 4π steradians. As shown in Fig. 2b, each module has a 60° feld of view in the elevation angle, and each one is Instrument design divided into fve channels, such that one channel corre- HEP consists of two types of telescopes, namely HEP-L sponds to 12°. HEP-L and HEP-H consist of three pin- and HEP-H, which have diferent geometric factors hole cameras, and each camera consists of a mechanical (G-factor) and energy ranges. As summarized in Table 1, collimator, stacked silicon semiconductor detectors, and HEP-L observes 70 keV–1.0 MeV electrons, and the application-specifc integrated circuits (ASICs) to read- −4 2 G-factor of its three detector modules is 9.3 × 10 ­cm out the signal. HEP-H has a larger collimator opening sr. HEP-H observes 0.7–2.0 MeV, and its G-factor is angle in order to attain a larger G-factor than that of −3 2 9.3 × 10 ­cm sr. Tese G-factors are numerically calcu- HEP-L. Additionally, HEP-H utilizes more detectors to lated and account only for the geometrical information detect higher-energy electrons. Because the electrode of regarding the collimator and detectors. Since the efec- the silicon detector is segmented, the position at which tive area of the detectors can be selected via parameters an incident electron interacts can only be determined in

Table 1 HEP performance and specifcations. Resolution is indicated by full-width at half maximum (FWHM) Parameter HEP-L HEP-H

Energy range 70 keV–1.0 MeV 0.7–2.0 MeV Energy resolution (ΔE/E) 11% at 300 keV, 18% at 750 keV 17% at 750 keV, 12% at 1.2 MeV Energy binning of onboard 70, 100, 124, 153, 188, 230, 280 340, 412, 499, 605, 730, 850, 990, 100, 153, 230, 340, 412, 499, 605, 730, 850, 990, ­histograma 1400, and 1800 (keV) 1200, 1400, 1600, 1800, and 2000 (keV) Field of view 10° (Azimuth) 60° (Elevation) for a module 10° (Azimuth) × 180° (Elevation) for three modules × Angular resolution 4° 1° 15° 3° ± 4 2 ± 3 2 Geometric factor 3.1 10− ­cm sr (one module) 3.1 10− ­cm sr (one module) × 4 2 × 3 2 9.3 10− ­cm sr (three modules) 9.3 10− ­cm sr (three modules) × × Time resolution 8 s per full 3-D distribution function (for normal spin period of 8 s) (15 histograms are generated for a 1/16th of spin period) Flux dynamic range 104–107/cm2/s/sr 103–106/cm2/s/sr Mass 6.8 kg Power 17.6 W a Hatched values are for calibration purposes. All values can be changed as onboard software parameters with commands issued from the ground Mitani et al. Earth, Planets and Space (2018) 70:77 Page 3 of 14

reference voltage are controlled by the control boards. HEP also has an interface for the software wave-particle interaction analyzer (S-WPIA) clock signal, which is used to determine the electron incidence timing and generate packets for the wave-particle analyzer software included in the mission data processor (Hikishima et al. 2018). We describe each of these functional blocks in further detail below.

Stacked silicon strip detector module Te total silicon thicknesses for the SSDs in HEP-L and HEP-H are 1.85 and 4.25 mm, respectively. Tese sili- con thicknesses correspond to the range of 850 keV and 1.7 MeV electrons, respectively, according to the ESTAR web database of electron stopping powers and ranges, which is maintained by the National Institutes of Stand- ards and Technology (NIST). Te incident direction of the detected particles is determined from the position of the interaction at the frst layer and the geometrical position of the collimator. To determine the interaction position, HEP uses SSDs, whose electrodes are sub- Fig. 1 Photograph of HEP fight model. The marked axes are the divided into closely spaced strips. As summarized in satellite coordinates shown in Fig. 2. In the left part of the fgure, it Table 2, HEP-L consists of one 50-μm-thick SSD and can be seen that the hexagonal cylinder houses three HEP-L modules three 600-μm-thick SSDs, while HEP-H consists of one and three HEP-H modules. Electronics boards are housed in the black 50-μm-thick SSD and seven 600-μm-thick SSDs. All box shown in the right part of the fgure. Each of the three sides of the hexagonal cylinder includes two cutouts through which the SSDs were manufactured by HAMAMATSU Photonics. electrons enter the HEP-L/HEP-H sensors K. K. Te full depletion voltage of the silicon wafers, from which the detectors were made, was 18 and 80 V for the 50- and 600-μm-thick SSDs, respectively. To choose the one dimension. Each HEP-L module utilizes four silicon operation bias voltages for the SSDs, we measured the strip detectors (SSDs), while each HEP-H module has spectra by using a radioisotope while changing the bias eight SSDs. Te collimator design and SSD placement are voltage. We chose 20 and 200 V as the operation bias illustrated in Fig. 3, and Table 2 lists the sizes of the SSDs. voltage for the 50- and 600-μm-thick SSDs, respectively. To block the protons, an Al sheet is placed in front of Signal processing the 50-μm-thick SSD. Te Al shield for HEP-L is 12.5 μm Electrons passing through the collimator interact with thick. Tis is the typical length traveled by a 0.9-MeV pro- the SSDs and deposit energy inside them. Te interaction ton in Al, based on the PSTAR proton-range table pro- positions and amount of deposited energy are measured vided by NIST. Terefore, the protons with lower energy by the SSDs in conjunction with the readout electron- are stopped in the Al sheet. To reduce the incidence of ics. A functional block diagram for the instrument is low-energy electrons in HEP-H, a 300-μm-thick Al sheet shown in Fig. 4. Te SSD modules in HEP-L and HEP-H is placed in front of the stacked SSDs. Tis thickness cor- are independently controlled by two control boards that responds to the typical distance traveled by the 6.5-MeV include feld programmable gate arrays (FPGAs) and protons or the 0.3-MeV electrons in Al. Te 50-μm-thick point-of-load (POL) DC–DC converters to power the SSD is also utilized to separate the electrons and pro- ASICs. Te central processing unit (CPU) board is part tons penetrating the Al sheet. To illustrate this concept, of the ERG mission network (Takashima et al. 2018) and Fig. 5 shows the results of a simple detector simulation utilizes the SpaceWire data link to receive command with the Geant4 toolkit, which simulates the passage of packets and send telemetry data. Te CPU board collects particles through matter. Figure 5a shows the diferen- the histogram data from the control boards, in addition tial fux of the electrons and protons at L = 3 based on to the data collected to monitor the instrument’s con- the AE8MAX/AP8MAX model (see the web page ‘AE-8/ dition. Te SSD bias voltage is supplied by a high-volt- AP-8 Radiation Belt Models’). Te simulation irradiates age–power supply board that includes two independent the electrons or protons according to the energy dis- high-voltage DC–DC converters, whose power and tribution onto the stacked SSD shielded by an Al sheet Mitani et al. Earth, Planets and Space (2018) 70:77 Page 4 of 14

a b Incident electron +Zsc

+Zsc +Ysc YZ-plane HEP-H HEP-L +Xsc A zim ut 0 ha Spacecraft 1 l c h a 2 n Incident θ n 3 e l electron θ 4 s SSD#1 5 6

XY-plane 7 +Ysc +Ysc SSD#3 8 +Xsc 9 Spacecraft SSD#2 HEP-H 10° HEP 10 10° HEP-L 11 θ: Elevation YZ-plane 12 14 13 +Zsc 60°

+Ysc HEP Fig. 2 a HEP mounted on ERG satellite and b instruments’ feld of view. The HEP instrument located on the X panel of the ERG satellite + sc (not to scale). The satellite rotates around the Zsc-axis. Fields of view of six detector modules are shown as gray areas; each module has a 60° (Elevation) 10° (Azimuth) feld of view. b Fields of view of three detector modules are shown as gray areas in the YZ-plane. Each module has a feld × of view of 60° and is divided into fve channels such that each channel covered 12°. Elevation angle (θ) is the angle between Z and the electron + sc incident direction

with the same thickness as that of the Al sheet installed HEP-H. Te standard deviations of the strip ID in HEP-L, on HEP-L. Figure 5b shows the simulated energy spectra which detects the maximum energy in the frst layer, are to be measured by the 50-μm-thick SSD. In this case, the 8.2 strips for the 100-keV electrons and 2.9 strips for the counts above the energy of 140 keV are generally caused 500-keV electrons, while those in HEP-H are 7.5 strips by protons. Terefore, proton contamination can be for the 750-keV electrons and 7.1 strips for the 1.5-MeV reduced by ignoring the events for which the energy at electrons. When calculated by the distance between the the thin SSD is larger than 140 keV. However, the energy collimator and the detector, which is indicated by R in threshold depends on the relative distributions of elec- Fig. 3, these values correspond to the incidence angles of trons and protons. By taking these considerations into 9.7°, 3.5°, 7.9°, and 7.4°, respectively. account, HEP is equipped with a function that assessed Te detection signal from each strip is read out by an an incident particle as a proton if the deposited energy at ASIC called VATA460.3. We have developed VATA460.3 the frst layer is higher than a certain threshold and if the in collaboration with IDEAS in Norway. Its development energy deposited in the second layer approximated zero. has carried out based on the designs of high-energy par- Tis function can be disabled, and the threshold energy ticle sensors aboard the BepiColombo MMO (Saito et al. Eth can be set remotely. 2010) and the soft gamma-ray detector aboard the Astro- Te position resolution is determined by the pitch H (Watanabe et al. 2014). As shown in Fig. 6, VATA460.3 of the strips in the 50-μm-thick SSD. To estimate the consists of 32-channel low-noise MOS amplifers and number of strips that will detect an incident charge, we Wilkinson 10-bit analog-to-digital converters (ADCs). simulate the interaction of electrons with the detector Te ASIC’s power consumption is 0.6 mW/ch. Each by using the Geant4 toolkit. In the simulation, a mono- channel included a charge-sensitive preamplifer, a fast energetic pencil beam of electrons is irradiated onto the CR-RC shaping amplifer followed by a discriminator that detector at normal incidence from the collimator input. generated a trigger signal for the readout sequence, and a 100- and 500-keV beams are simulated to test HEP-L, slow CR-RC shaping amplifer followed by a sample-and- and 750-keV and 1.5-MeV beams are simulated to test hold circuit for pulse-height analysis. Te preamplifers Mitani et al. Earth, Planets and Space (2018) 70:77 Page 5 of 14

Top view HEP-L HEP-H A 0.63 1.70 B 8.98 33.43 B b C 23.0425.60 a 0.63 2.40 b 0.63 2.40 L Collimator c 3.53 5.53 L 7.2 27.5 R 9.5 10.9 R A a d 5.3 12.5 d SSDs (All units in mm. The manufacturing errors for C and c are less than 0.001 mm) C c

Dimensions of square holes in collimator plates HEP-LHEP-H Front view Side view xthickness xthickness 8.98 0.3 33.43 1.5 8.58 0.3 31.66 1.0 8.22 0.3 30.46 1.0 7.86 0.3 29.26 1.0 7.50 0.3 28.06 1.0 7.14 0.3 26.86 1.0 6.78 0.3 25.67 1.0 6.41 0.3 24.47 1.0 Top view 6.05 0.3 23.27 1.0 5.69 0.3 22.07 1.0 5.33 0.3 20.87 1.0 B b 4.97 0.3 19.67 1.0 4.61 0.3 18.48 1.0 4.25 0.3 17.28 1.0 3.88 0.3 16.08 1.0 3.52 0.3 14.88 1.0 3.16 0.3 13.68 1.0 2.80 0.3 12.48 1.0 2.44 0.3 11.29 1.0 L Collimator 2.08 0.3 10.09 1.0 1.71 0.38.891.0 1.35 0.37.691.0 0.99 0.36.491.0 0.63 0.35.301.0 a 4.10 1.0 R A 2.90 1.0 1.70 1.0

(All units in mm. The manufacturing errors of x and the thickness are 0.03 mm and 10 %, respectively. ) d SSDs

C c

Front view Side view Fig. 3 Collimator geometry and size of detector in one HEP module. Upper fgure illustrates collimator and stack of silicon strip detectors (SSDs) (one 50-μm-thick detector and three 600-μm-thick detectors). Lower fgure marks dimensions of collimator and detector stack of HEP-H. The silicon stack of HEP-H consists of one 50-μm-thick sheet and seven 600-μm-thick sheets. The values of A, B, C, a, b, c, L, R, and d are listed in the table at the upper right area of the fgure. When not otherwise noted, the manufacturing tolerance is 0.1 mm. HEP-L and HEP-H collimators consist of stacked thin Al plates with square holes, and their dimensions are listed in the lower right table Mitani et al. Earth, Planets and Space (2018) 70:77 Page 6 of 14

Table 2 Sizes of silicon strip detectors used in HEP Layer (µm) Thickness (µm) Strip pitch­ a (µm) Strip length­ a (mm) Number of strips

HEP-L Al sheet 12.5 2.5 ± 1st layer 50 10 200 3.53 67 ± 2nd–4th layer 600 10 360 3.53 64 ± HEP-H Al sheet 300 30 ± 1st layer 50 10 200 5.53 78 ± 2nd–8th layer 600 10 400 5.53 64 ± a The manufacturing errors for the strip pitch and the strip length are less than 1 µm ± accept a maximum input of 90 fC, which corresponds each. Terefore, 9 and 17 ASICs are used in one SSD to 2 MeV in Si. Te fast and slow shaping times are 0.6 module of the HEP-L and HEP-H, respectively. To tune and 2 µs, respectively, and each one can be tuned by the bias parameters and channel logic, each ASIC has a using the control registers. Te Wilkinson ADC con- 520-bit control register. Te tunable parameters include verts the held pulse height to a digital value in less than the shaping time and global threshold for each chip, and 100 µs. Te digital data from the 32 channels are mul- information on whether the trigger output for each chan- tiplexed and displayed on a serialized digital interface. nel is enabled. With these registers, we can change which Te strips in the frst layer are read out by three ASICs, channels issues the trigger signals from the ground. Tis and those in the lower layers are read out by two ASICs means that the efective detection area can be reduced

Fig. 4 Functional block diagram of HEP instrument. The three HEP-L and HEP-H SSD modules are controlled independently by two control boards. A central processing unit (CPU) board receives command packets and sends telemetry data to the mission network. The CPU board collects histogram data from the control boards in addition to instrument-status data. The bias voltage of the SSDs is supplied from a high-voltage–power supply board that includes two independent high-voltage DC–DC convertors, with the power and reference voltage controlled by the control boards Mitani et al. Earth, Planets and Space (2018) 70:77 Page 7 of 14

a b

104

3 10

2

Counts 10

10

1 100150 200 250 300 350 400 450 500 Detected energy [keV] Fig. 5 a Diferential fuxes of electrons and protons at L 3 based on AE8 MAX and AP8 MAX models; b detected energy spectra in frst layer of = stacked SSD when fux shown in a irradiates stacked SSDs of HEP-L

Fig. 6 Functional block diagram for VATA460.3, which consists of 32-channel low-noise MOS amplifers and Wilkinson 10-bit analog-to-digital converters (ADCs). Each channel includes a charge-sensitive preamplifer, a fast CR-RC shaping amplifer followed by a discriminator generating a trigger signal for the readout sequence, and a slow CR-RC shaping amplifer followed by a sample-and-hold circuit for pulse-height analysis. The digital data from the 32 channels are multiplexed and read out on a serialized digital interface Mitani et al. Earth, Planets and Space (2018) 70:77 Page 8 of 14

by disabling the triggers, which resulted to a reduced larger than the threshold. Te control board sends a sam- G-factor. ple/hold signal to all of the ASICs with a defned delay after the triggering signal. Te control board supplies Data processing and observation modes the ADC clocks and receives a signal from the ASICs When the control board receives a trigger signal from when the conversion has fnished. After receiving these one of the SSD modules, it issues a sample/hold signal signals from all of the ASICs, the control board supplied to the ASICs and has them convert the analog data to the readout clocks of the ASICs. After reading out all of digital data. Ten, the control board reads out the digi- the data, the control board resets the digital part of the tal data from all of the ASICs. Subsequently, the control ASICs and waits for the next trigger signal. Te total board converts the ADC outputs to energy values and deposited energy and incident direction for each electron adds the contributions from all layers in order to calcu- event are determined in this sequence. HEP generates late the total deposited energy. Te detailed signal pro- telemetry data according to the operation mode and has cessing of an event is illustrated in Fig. 7 and is as follows: the following four operation modes: setting mode, nor- a trigger signal is issued if more than one of the outputs mal observation mode, S-WPIA observation mode, ASIC from the fast shapers in the ASICs of the frst SSD layer is calibration mode, and event mode. In the setting mode,

Fig. 7 Timing chart for signal processing with VATA460.3. Readout sequence begins with trigger signal from ASICs in frst SSD layer. The CONTROL board inputs a sample/hold signal to all ASICs after a defned delay from the trigger signal. The CONTROL board supplies the ADC clocks and receives a signal from the ASICs when the conversion is fnished. After receiving confrmation from all ASICs that the A/D conversion has completed, the CONTROL board supplies the ASICs readout clock. After reading out all the data, the CONTROL board resets the digital part of the ASICs and waits for the next trigger signal Mitani et al. Earth, Planets and Space (2018) 70:77 Page 9 of 14

the ASIC control parameters, FPGA registers, and CPU software parameters can be set. Moreover, the ASIC con- 120 60 keV -ray trol registers can only be set in this mode. In the normal 13.9 keV 17.6 keV γ observation mode, the energy histograms are generated 100 onboard. Fifteen histograms corresponding to the chan- 80 nels shown in Fig. 2 are generated based on the deposit 60 energy and incident elevation angle of each detected Counts particle (θ in Fig. 2). Te data accumulation continues 40 for 1/16th of the satellite spin period, and the data are 20 sent by the CPU software as telemetry data. Terefore, 0 in normal observation mode, 240 histograms are gener- 010203040506070 ated for every spin. Each histogram has 16 energy bins, ADC channel as listed in Table 1. In the S-WPIA observation mode, Fig. 8 γ-ray spectrum obtained by HEP SSD recorded with 241Am the packets for the S-WPIA are produced in addition radioactive isotope to the normal observation mode data. Te energy, inci- dent direction, and timing information of each electron are sent to the mission data recorder. Te prime objec- calculated in order to serve as the conversion factor of tive of the S-WPIA is to measure the energy exchanged the ADC to return energy values. Figure 9 shows the between the whistler-mode chorus emissions and the gains of all ASIC channels. HEP-L and HEP-H have 864 energetic electrons in the inner magnetosphere (Katoh and 1632 channels, respectively (nine ASICs per HEP-L et al. 2018). By assuming that 10 kHz is the highest elec- module and 17 ASICs per HEP-H module). Channels tron cyclotron frequency along the satellite’s orbit at the 0–95, 288–384, and 576–672 for HEP-L, and channels equator, the wave period of the chorus emissions is found 0–95, 544–640, and 1088–1184 for HEP-H correspond to be approximately 100 μs. A detection accuracy greater to the 50-μm-thick SSD in the frst layer of the stacked than 10 μs can resolve the wave phase in the order of detectors. Because the deposited energy is expected to tens of degrees. To achieve this accuracy, HEP receives be small in the 50-μm-thick SSD, the parameters to con- the S-WPIA clock signals at 524.288 kHz (1.9-μs period) trol the ASIC A/D conversion are set such as needed to from the plasma wave experiment (Kasahara et al. 2018b) obtain a higher ADC value than that of the 600-μm-thick instrument and has a counter module to count the clock SSD. Te variation of the gains is caused by the individual signals. When HEP receives a trigger signal from the ASIC characteristics. In the frst layer, the average gain of SSD modules, it latches the counter and adds the value the ASICs is 0.92 ch/keV, while that in the lower layers is to the S-WPIA data in order to achieve time indexing at 0.46 ch/keV. Additionally, the standard deviations are 4.5 a resolution of 1.9 μs. In the ASIC calibration mode, the and 2.9%, respectively. With these settings, the ASICs of waveforms of all ASIC channel shaping amplifer outputs the frst layer and those of the lower layers cover the inci- are produced with test pulses. Te raw ASIC data can dent energies up to 1 and 2 MeV, respectively. Te SSD be obtained in the event mode, and all individual event dead layer is less than 3 μm, where the 70-keV electrons ASIC channel data are sent as telemetry data, for detailed lose 2.9 keV (according to ESTAR). For the HEP require- calibration purposes. However, only a small portion of ments, this efect can be ignored. the triggering data can be relayed because the available Te overall performance of the HEP was tested with telemetry data are limited. electron beams from a particle accelerator, and a single- In accordance with the ERG science observation plans, ended Pelletron system (National Electrostatics Cor- in the normal observation phase of ERG, HEP operates poration, Model 6SH) at JAXA’s Tsukuba Space Center in the normal observation mode by default and enters (see the webpage of Space environment measurement the S-WPIA mode several times during a one-orbit laboratory at JAXA). Te beam intensity ranges from 1 revolution. fA to 10 nA, and the system can generate electrons with energies of 0.4–2.0 MeV. Te HEP was placed in a vac- Prefight testing uum chamber such that the beam direction was parallel For the energy calibration of all SSD channels, we tested to the YscZsc-plane marked in Fig. 1. Ten, it was irradi- the instrument with radioactive isotopes. We measured ated with monoenergetic electron beams. By rotating the energy spectra of all strips by using γ-ray sources the HEP instrument around the Xsc-axis in the chamber, placed in front of each SSD module. Figure 8 shows the all SSD modules were irradiated with the beam and the spectrum of one SSD channel. From the center of the detected energies and angular responses were recorded. 60-keV γ-ray peak shown in this fgure, the gain was Te beam intensity was tuned such that the count rates Mitani et al. Earth, Planets and Space (2018) 70:77 Page 10 of 14

1.2 SSD #1 SSD #2 SSD #3 1

0.8 Layer-1 -2 -3 -4 0.6

0.4 Gain [ ch/keV ] 0.2 a HEP-L

0 0 100 200 300 400 500 600 700 800

1.2 SSD #1 SSD #2 SSD #3 1

0.8 Layer -1 -2 -3 -4 -5 -6 -7 -8 0.6

0.4 Gain [ ch/keV ] 0.2 b HEP-H

0 0 200 400 600 800 1000 1200 1400 1600 ASIC channel Fig. 9 VATA460.3 gain of all a HEP-L and b HEP-H channels of SSD modules. The dotted vertical lines indicate the frst channel of each ASIC. The strips in the frst layer are read out with three ASICs, and those in the lower layer are read out with two ASICs per layer. Therefore, 9 and 17 ASICs are used in the HEP-L and HEP-H SSD modules, respectively. Channels 0–95, 288–384, and 576–672 of HEP-L, and channels 0–95, 544–640, and 1088–1184 of HEP-H correspond to the frst layer, which is the 50-μm-thick SSD

abHEP-L HEP-H 300

300 keV 500 keV 750 keV 750 keV 1.2 MeV 250 250

200 200

125 keV 143 keV 150 150 1.8 MeV Counts Counts 89 keV 118 keV 34 keV 100 100 231 keV

50 50

0 0 0200 400 600 800 1000 0 500 1000 1500 2000 Detected energy [keV] Detected energy [keV] Fig. 10 Electron spectra measured with HEP fight model. a Spectra detected by HEP-L show electrons with 300, 500, and 750 keV of energy. b Spectra detected by HEP-H show electrons with energy of 750 keV, 1.2 MeV, and 1.8 MeV. Dotted line shows result of ftting with Gaussian function in addition to straight line Mitani et al. Earth, Planets and Space (2018) 70:77 Page 11 of 14

for the SSD module were approximately a few hundred HEP-H, respectively. To evaluate the angular resolution, Hz. Figure 10 shows the energy spectra obtained with the histograms had more bins than the azimuthal chan- the HEP fight model when the input electron energy nels in Fig. 2, while the histograms of HEP-L and HEP-H was 300, 500, and 750 keV for HEP-L, and 750 keV, had 60 bins and 12 bins covering 60°, respectively. Tus, 1.2 MeV, and 1.8 MeV for HEP-H. In these tests, we the bin width of the HEP-L and HEP-H responses were used the electron beam with a normal incidence to the 1° and 3°, respectively. Tese values are smaller than SSDs. Te triggers from the strips in the frst layer were the angle uncertainty caused by the position resolution, enabled, and the total deposited energy was determined which is estimated based on the detector simulation by summing up the contributions of all layers. Te spec- described above. Based on the responses, the angular res- tral peaks were ftted with a Gaussian function in addi- olution had a FWHM of approximately 4° (4 bins) and 15° tion to a straight line. Te full-width at half maximum (5 bins) for HEP-L and HEP-H, respectively. (FWHM) values were 34 keV at 300 keV, 89 keV at 500 keV, and 118 keV at 750 keV for HEP-L, while those In‑orbit operation and fight performance for HEP-H were 125 keV at 750 keV, 143 keV at 1.2 MeV, On February 2, 2017, HEP was turned on for the frst and 231 keV at 1.8 MeV. An energy resolution of less than time while in orbit, and the initial checkout was success- 20% was realized. According to ESTAR, the energy loss ful. Te bias voltage of the silicon detectors was limited at the 300-μm-thick Al sheet of HEP-H is 0.12 MeV for a below 50 V, and the count rates were monitored for sev- 0.7–2 MeV electron. Te diference between the energy eral days. of the input electrons and the detected energy was con- On February 6, 2017, a nominal bias voltage of 200 V sistent with the energy loss. Moreover, the loss should was applied, and the detector performance was tested be considered when we infer the incident energy of the and found to be normal. We checked the noise level and electrons from the detected signals, and the low-energy waveform of the shaping amplifer output for every chan- tails in the spectra should also be considered. To evalu- nel. Te waveforms of several channels are shown in ate the incident-angle response, the HEP instrument was Fig. 12, alongside the waveforms measured before launch, rotated around the Xsc-axis marked in Fig. 1. Te meas- for comparison purposes. Te peak times and pulse ured angular responses when changing the beam inci- heights were not diferent before and after the launch. dent angle are shown in Fig. 11. Te fgure shows the After the other instruments aboard the ERG had fn- histograms that resulted from the measurements with ished the initial checkouts followed by the verifcation of seven and three diferent incident angles for HEP-L and the overall operation plan, the ERG satellite was shifted

abHEP-L HEP-H 45

40 30

35 25 30 20 25 Count s Count s 20 15

15 10 10 5 5

0 0 0102030405060 0102030405060 Elevation [deg] Elevation [deg] Fig. 11 Angular response measured with HEP fight model. a Angular responses measured while changing the beam incident angle on HEP-L (i.e.,

rotating HEP around Xsc-axis in Fig. 1); bin width is 1°. b Angular responses measured when changing beam incident angle on HEP-H (i.e., rotating HEP around Xsc-axis in Fig. 1); bin width is 3° Mitani et al. Earth, Planets and Space (2018) 70:77 Page 12 of 14

SSD#1 ASIC-3 ch 0 - 15 SSD#1 ASIC-3 ch 0 - 15 1000 ASIC-3 1000 ASIC-3 HEP-L ( before launch ) ch 0 ch 1 HEP-L ( in-orbit ) ch 0 ch 1 ch 2 ch 3 ch 2 ch 3 ch 4 ch 5 ch 4 ch 5 800 ch 6 ch 7 800 ch 6 ch 7 ch 8 ch 9 ch 8 ch 9 ch 10 ch 11 ch 10 ch 11 ch 12 ch 13 ch 12 ch 13 ch 14 ch 15 ch 14 ch 15 600 600

400 400 ADC channel ADC channel

200 200

0 0 012345678 012345678 Hold delay [µs] Hold delay [µs]

SSD#1 ASIC-3 ch 0 - 15 SSD#1 ASIC-3 ch 0 - 15 1000 ASIC-3 1000 ASIC-3 HEP-H ( before launch ) ch 0 ch 1 HEP-H ( in-orbit ) ch 0 ch 1 ch 2 ch 3 ch 2 ch 3 ch 4 ch 5 ch 4 ch 5 ch 6 ch 7 ch 6 ch 7 800 ch 8 ch 9 800 ch 8 ch 9 ch 10 ch 11 ch 10 ch 11 ch 12 ch 13 ch 12 ch 13 ch 14 ch 15 ch 14 ch 15 600 600

400 400 ADC channel ADC channel

200 200

0 0 012345678 012345678 Hold delay [µs] Hold delay [µs] Fig. 12 Waveforms of shaping amplifer output measured while in orbit (right) compared to prefight measurements (left) for performance verifcation to its normal observation phase in late March 2017. Since Earth’s inner magnetosphere. Te HEP consists of three March 16, 2017, HEP has been operating in the normal HEP-L modules and three HEP-H modules. HEP-L observation mode, and several parameters controlling detects 70 keV–1 MeV electrons and has a maximum −4 2 the channels triggered by the SSDs have been changed. G-factor of 9.3 × 10 ­cm sr, while HEP-H observes Figure 13 shows the energy–time (E–t) spectrograms, 0.7 MeV–2 MeV electrons and has a maximum G-fac- −3 2 which represent the change in the time of the spin- tor of 9.3 × 10 ­cm sr at maximum. Tese modules averaged count rate observed by HEP with the L-value, are pin-hole cameras consisting of mechanical collima- which is the radial distance of the electron drift orbits in tors and SSDs. Te signals from a total of 2355 strips are the magnetic equatorial plane, and is measured in units processed by 78 readout ASICs. Before the satellite was of Earth radii. Te E–t spectrograms in Fig. 13 cover launched, all channels were evaluated with reference sig- 2 days and fve orbital revolutions. As the satellite moves nals from radioactive isotopes and the overall HEP per- through diferent L-values, the observed count rates formance was evaluated with electron beams. In orbit, change and peak at L = 4–5, which corresponds to the the waveforms of the calibration pulses indicated that the outer radiation belt, as expected. Near the orbit’s perigee HEP functioned properly after it was launched. From the (L = 3), HEP ceases the observation because of the high initial results of the energy–time spectrograms, the HEP count rates of high-energy protons. Te counts measured recorded high electron count rates in the outer radiation with HEP-L in the vicinity of 0.9 MeV increase near the belt. perigee because of high-energy protons. A simulation of the detector is under development in order to convert the count data to physical quantities Summary and future work with higher precision. We will model the HEP detector Te HEP instrument has successfully begun the observa- geometry and particle interaction with the detectors and tion of electrons with energies of 70 keV–2 MeV in the surrounding materials by utilizing the Geant4 library. Mitani et al. Earth, Planets and Space (2018) 70:77 Page 13 of 14

7

6

5

4 Lm 3

[dimensionless] 2

1

0 102

101

H 100 1000 prov. [keV] HEP- Energy omni cnt

-1

10 [cnt/sample/keV]

10-2

102

1000 101

100 prov. [keV] HEP- L Energy omni cnt

-1

10 [cnt/sample/keV]

100 10-2

Lm 3.4 5.5 6.0 6.5 5.1 4.7 2.3 MLT 7.3 5.9 4.9 3.9 3.0 0.7 9.3 MLAT 8.6 10.5 -4.9 -17.3 -8.6 -38.6 20.8 hhmm 0000 0800 1600 0000 0800 1600 0000 2017 Mar 17 Mar 18 Mar 19 Fig. 13 Energy–time spectrograms observed via HEP fight-count data on March 17 and 18, 2017. The top panel shows L-value of ERG satellite position. Middle and bottom panels show energy–time spectrograms from HEP-L and HEP-H, respectively. Colors indicate raw count not corrected for G-factor or detection efciency

After the validation tests of the simulator and the com- Author details 1 Institute of Space and Astronautical Science, Japan Aerospace Explora- parisons between the simulation results and the experi- tion Agency, Sagamihara 252‑5210, Kanagawa, Japan. 2 Department mental results, we will be able to calculate the detector of Earth and Planetary Science, School of Science, The University of Tokyo, 3 response by using the simulator. As shown in Fig. 10, Bunkyo‑ku 113‑0033, Tokyo, Japan. Department of Aeronautics and Astro- nautics, Tokai University, Hiratsuka‑shi 259‑1292, Kanagawa, Japan. 4 Institute monoenergetic electrons can be detected in lower energy of Space‑Earth Environmental Research, Nagoya University, Nagoya 464‑8601, channels. A spectrum detected in orbit is the superposi- Aichi, Japan. tion of signals from electrons with diferent energy. By Acknowledgements using the simulator, we can estimate how many of the We would like to thank all members of the ERG project for their long-lasting detected counts in the lower energy channels are contrib- eforts to realize this mission. The HEP instrument was developed by Mitsubi- uted by higher-energy electrons, and thereby determine shi Heavy Industries Co, Ltd. The high-voltage board in HEP was developed by Meisei Electric Co., Ltd. The silicon detectors were manufactured by Hama- the incident fux with higher precision. Tus, we will be matsu Photonics K. K., and VATA460.3 was manufactured by IDEAS, in Norway. able to deduce the distribution of incident electrons with The development of VATA460.3 has profted from the design heritage of higher precision from the direction and energy detec- BepiColombo MMO and Astro-H SGD/HXI. From the Astro-H team, we would like to thank S. Watanabe and M. Kokubun in particular, for their support and tions in orbit. A detailed report regarding the simulator sharing of experience. The electron beam facility used in the calibration of and its validation will be published in the future. HEP is maintained by the Research and Development Directorate in JAXA. Quick plots are distributed by the ERG science center, which also makes Authors’ contributions level-2 data available. We also thank F. Makino, T. Goka, H. Tajima, and T. Mukai, TM led the design and development of HEP. TT supported and ofered advice who reviewed the design and development of HEP during each phase of its during all phases of design and development. SK provided information neces- development. sary to the development with regard to the instruments of the ERG mission. WM worked on the basic development of the VATA460 series. MH worked on Competing interests the initial collimator design and stacked SSD module. All authors read and The authors declare that they have no competing interests. approved the fnal manuscript. Mitani et al. Earth, Planets and Space (2018) 70:77 Page 14 of 14

Availability of data and materials Kazama Y, Wang BJ, Wang SY, Ho PTP, Tam SWY (2017) Low-energy particle The data and materials used in this research are available from the corre- experiments—electron analyzer onboard the Arase spacecraft. Earth sponding author upon reasonable request. Planets Space 69:165. https​://doi.org/10.1186/s4062​3-017-0748-6 Makishima K (1999) Energy non-equipartition processes in the Universe. Consent for publication Astron Nachr 320:163–166 Not applicable. Miyoshi Y, Ono T, Takashima T, Asamura K, Hirahara M, Kasaba Y, Matsuoka A, Kojima H, Shiokawa K, Seki K, Fujimoto M, Nagatsuma T, Cheng CZ, Ethics approval and consent to participate Kazama Y, Kasahara S, Mitani T, Matsumoto H, Higashio N, Kumamoto A, Not applicable. Yagitani S, Kasahara Y, Ishisaka K, Blomberg L, Fujimoto A, Katoh Y, Ebihara Y, Omura Y, Nose Hori T, Miyashita Y, Tanaka Y-M, Segawa TT, ERG working Funding group (2012) The Energization and Radiation in Geospace (ERG) project. The ERG (Arase) project is funded by ISAS/JAXA. In: Summers D, Mann IR, Baker DN, Schulz M (eds) Dynamics of the earth’s radiation belts and inner magnetosphere. American Geophysical Union, Washington. https​://doi.org/10.1029/2012G​M0013​04 Publisher’s Note Nagai T (1988) Space weather forecast: prediction of relativistic electron Springer Nature remains neutral with regard to jurisdictional claims in pub- intensity at synchronous orbit. Geophys Res Lett 15:425. https​://doi. lished maps and institutional afliations. org/10.1029/GL015​i005p​00425​ Reeves GD, McAdams KL, Friedel RHW, O’Brien TP (2003) Acceleration and loss Received: 4 October 2017 Accepted: 25 April 2018 of relativistic electrons during geomagnetic storms. Res Lett, Geophys. https​://doi.org/10.1029/2002G​L0165​13 Takashima T, Ogawa E, Asamura K, Hikishima M (2018) Design of a mission network system using SpaceWire for scientifc payloads onboard the Arase spacecraft. Earth Planets Space 70:71. https​://doi.org/10.1186/ References s4062​3-018-0839-z Baker DN, Blake JB, Klebesadel RW, Higbie PR (1986) Highly relativistic Saito Y, Sauvaud JA, Hirahara M, Barabash S, Delcourt D, Takashima T, Asamura electrons in the earth’s outer magnetosphere: 1. Lifetimes and temporal K, BepiColombo MMO, Team MPPE (2010) Scientifc objectives and instru- history 1979–1984. J Geophys Res 91(A4):4265–4276. https​://doi. mentation of Mercury Plasma Particle Experiment (MPPE) onboard MMO. org/10.1029/JA091​iA04p​04265​ Planet Space Sci 58:82–200. https​://doi.org/10.1016/j.pss.2008.06.003 Hikishima M, Kojima H, Katoh Y, Kasahara Y, Kasahara S, Mitani T, Higashio N, Watanabe S, Tajima H, Fukazawa Y et al (2014) The Si/CdTe semiconductor Matsuoka A, Miyoshi Y, Asamura K, Takashima T, Yokota S, Kitahara M, Compton camera of the ASTRO-H Soft Gamma-ray Detector (SGD). Matsuda S (2018) Data processing in the software-type wave-particle Nucl Instrum Methods Phys Res A 765:192. https​://doi.org/10.1016/j. interaction analyzer on board the Arase satellite. Earth Planets Space nima.2014.05.127 70:80. https​://doi.org/10.1186/s4062​3-018-0856-y Katoh Y, Kojima H, Hikishima M, Takashima T, Asamura K, Miyoshi Y, Kasahara Y, Kasahara S, Mitani T, Higashio N, Matsuoka A, Ozaki M, Yagitani S, Yokota Online databases S, Matsuda S, Kitahara M, Shinohara I (2018) Software-type wave-particle AE-8/AP-8 Radiation Belt Models. https​://ccmc.gsfc.nasa.gov/model​web/ interaction analyzer on board the Arase satellite. Earth Planets Space 70:4. model​s/trap.php https​://doi.org/10.1186/s4062​3-017-0771-7 Geant4 toolkit. http://geant​4.cern.ch Kasahara S, Yokota S, Mitani T, Asamura K, Hirahara M, Shibano Y, Takashima T Space environment measurement laboratory, JAXA. http://sees.tksc.jaxa.jp/ (2018a) Medium-energy particle experiments—electron analyser (MEP-e) fw_e/dfw/SEES/Engli​sh/Labo/labo_e.shtml​ for the energization and radiation in geospace (ERG) mission. Earth Plan- Stopping-Power & Range Tables for Electrons, Protons, and Helium Ions, ets Space 70:69. https​://doi.org/10.1186/s4062​3-018-0847-z National Institute of Standards and Technology, U.S. Department of Com- Kasahara Y, Kasaba Y, Kojima H, Yagitani S, Ishisaka K, Kumamoto A, Tsuchiya merce. https​://www.nist.gov/pml/stopp​ing-power​-range​-table​s-elect​ F, Ozaki M, Matsuda S, Imachi T, Miyoshi Y, Hikishima M, Katoh Y, Ota M, rons-proto​ns-and-heliu​m-ions Shoji M, Matsuoka A, Shinohara I (2018b) The plasma wave experiment (PWE) on board the Arase (ERG) satellite. Earth Planets Space 70:86. https​ ://doi.org/10.1186/s4062​3-018-0842-4