Measurement by FIB on the ISS: Two Emissions of Solar Neutrons Detected?

A new type of solar neutron detector (FIB) was launched onboard the Space Shuttle Endeavour on July 16, 2009, and it began collecting data at the International Space Station (ISS) on August 25, 2009. This paper summarizes the three years of observations obtained by the solar neutron detector FIB until the end of July 2012. The solar neutron detector FIB can determine both the energy and arrival direction of neutrons. We measured the energy spectra of background neutrons over the SAA region and elsewhere, and found the typical trigger rates to be 20 counts/sec and 0.22 counts/sec, respectively. It is possible to identify solar neutrons to within a level of 0.028 counts/sec, provided that directional information is applied. Solar neutrons were observed in association with the M-class solar flares that occurred on March 7 (M3.7) and June 7 (M2.5) of 2011. This marked the first time that neutrons were observed in M-class solar flares. A possible interpretaion of the prodcution process is provided.


Introduction -a brief history of solar neutron detection
High-energy protons coming from the Sun on February 28 and March 3, 1942 were first discovered by Forbush and published in 1946. [1] In 1951, Biermann, Haxel, and Schulter had predicted the potential discovery of solar neutrons on Earth. [2] Neutrons are produced when the accelerated ions strike the solar surface. However, solar neutrons were actually detected 29 years after this prediction. A clear signal of gamma rays and neutrons was detected in association with a large solar flare on June 21, 1980, with an X-ray intensity of X2.5, by the Gamma Ray Spectrometer composed of the NaI and CsI detectors on board the Solar Maximum Mission (SMM) satellite. [3] Figure 1 shows the results. The first peak corresponds to the gamma-ray signal, while the second was induced by the neutron signal.
Because neutrons cannot travel from the Sun to Earth at the speed of light, their arrival time distribution is associated with a time delay from the speed of light, even when simultaneously released from the Sun. For the time distribution presented in Fig. 1, if the same departure time for neutrons is set, a neutron energy spectrum is obtained, which assumed an impulsive production of neutrons on the Sun. The spectrum can be expressed by a power law: En -γ dEn with γ = 3.5±0.1. [3] Two years later, on June 3, 1982, the SMM satellite again detected a neutron signal. [4] However, neutron monitors located on the ground have successively detected neutron signals in association with a large X8.2 solar flare [5] , which shed new light on the production time of neutrons in the solar atmosphere. One component involved in the data cannot be explained by an impulsive production mechanism alone. Only two solar neutron events had been accumulated until solar cycle 21, and it was too early to judge the production time of neutrons in the solar atmosphere, namely, whether high-energy neutrons are produced impulsively or gradually. Both scenarios were possible for the same event and this would be a great challenge for solar physicists.
To identify the production time of neutrons at the solar surface in the solar cycle 22, new detectors capable of measuring the energy of neutrons were expected. Therefore a new type of solar neutron detector -the solar neutron telescope (SONTEL) -was designed, based on the plastic scintillator.
SONTEL can measure the energy and direction of neutrons using the charge exchange process of neutrons into protons [6] . Therefore SONTEL can determine the flight time from the Sun to Earth.
Of course, conventional Simpson-type neutron monitors were also operated [7][8] . At the same time, the possibility of launching a new type of solar neutron detector into space was considered [9] . To resolve the mystery regarding the production time of neutrons, it is inevitable to have a new type of detector capable of measuring the energy of neutrons in space. We could then identify when those neutrons left the Sun.
An attempt to measure the energy of neutrons was found in a paper dated around 1985 [10] . Scintillator bars composed of two layers were equipped (in X and Y directions) and the device was circulated over the Southern hemisphere by a Racoon balloon flight from Alice Springs, Australia to detect solar neutrons within the energy range of 20 to 150 MeV, but detected no signals. Almost the same year 1985, another new instrument was proposed, capable of unambiguously determining the energy and direction of incident neutrons using a technique, the double Compton scattering method [11] . The detector was named SONTRAC. Independent from these activities, most of which were developed in the USA, in Japan, a detector comprising a mass of the scintillation fiber is proposed to detect anti-deuterium in space and neutron and anti-neutron oscillation in the space between the Sun and Earth. In 1991, we proposed a new type of detector for the Japanese Experimental Module (JEM) of the International Space Station (ISS) [12][13] . Gamma-rays accounted for the first peak; neutrons accounted for the second. The original picture was prepared by the authors of the ref [3].
In April 1991 and August 1991, a large gamma-ray satellite CGRO and a solar satellite Yohkoh were launched. An image of the Sun using neutron signals was successfully drawn, using the Compton scattering function of the COMPTEL detector [14] and beautiful photographs of solar flares were taken by using the soft and hard X-ray telescope of Yohkoh satellite. They have left very important archives on the solar activities [15] .
During the solar cycle 22, several new discoveries involving solar neutrons were made; based not only on many ground level detectors but also a few that were space-borne.
Consequently, more solar neutron events were accumulated, including one highlight, the discovery of an extremely strong signal of neutrons in association with the large X9.3 solar flare on May 24, 1990 [16] . The signal was the strongest ever observed by the neutron monitor. In association with this flare, two Soviet satellites, GRANAT/PHEBUS [17] and GAMMA-1 [18][19] successfully captured very impulsive high-energy gamma rays starting at 20:48 UT. One minute later (20:49UT), strong neutron signals were detected by many neutron monitors located throughout the North American continent [20] . Subsequently, from around 21:00UT, Ground Level Enhancement (GLE) was observed, induced by high-energy protons. The key knowledge obtained by the event on May 24, 1990, reviewing some 20 years after the discovery, may be the sudden increase in the ratio between 70-95MeV gamma rays and 4-7 MeV nuclear gamma rays 3 minutes later.
Chupp and Ryan summarized that the change in ratio may have been induced by accelerated protons to several hundred MeV [21] . It is worth noting that to detect high-energy neutrons at ground level, conventional neutron monitors were not only used to detect solar neutrons in the large solar flare on May 24, 1990, but also in the X9.4 flare on March 22, 1991 [22] .
The subsequent scope of remarkable events from solar cycle 22 may also include the detections of high-energy gamma rays and neutrons in association with the six extremely powerful solar flares with X12 observed during June 1 and 15, 1991. Solar neutrons were detected in association with two large solar flares on June 4 and 6, 1991, respectively, using two kinds of solar neutron detectors located on Mt. Norikura: the solar neutron telescope [23] and the neutron monitor [24] . Via simultaneous observations with the neutron monitor and the neutron telescope, the capability of the new solar neutron telescope was demonstrated.
It should be mentioned that in the solar flare on June 4, 1991, the BATSE [25] and OSSE [26] detectors on board the CGRO satellite observed the long-standing emission of gamma rays with a decay time of 330 seconds after a sharp impulse signal. OSSE observed a neutron capture line (2.223 MeV) and a carbon de-excitation line (4.44 MeV) that continued for three hours. High-energy gamma rays were detected by the EGRET detector with energies of 50 to 100 MeV and >150 MeV for the flare events on June 4, 6, 9, and 11. Moreover, in the flare event on June 11, a particularly long-lasting emission of high-energy gamma rays was recorded, lasting 10 hours [27] .
Many arguments concerning the long-lasting gamma rays emerged at the time, namely whether they were induced by the continuous acceleration process of protons (such as in the shock acceleration model [28] ) or by protons trapped in the magnetic loop and precipitating on the solar surface. [29][30] The impulsive production mechanism of neutrons on the solar surface was attributable to the reconnection process of magnetic loops [31][32][33] or the DC acceleration mechanism [34] , while long-lasting emissions of gamma rays may be closely related to the shock acceleration process. The question of whether the long-lasting high-energy gamma-ray emission is attributable to the continuous acceleration of the protons above 300 MeV [35] , or the injection of flare accelerated particles into a large coronal loop with release at the mirror points of the loop where the gamma rays are produced, is very interesting [29][30] , the final answer to which will hopefully be obtained in solar cycle 24.
During the solar flare event on September 7, 2005, solar neutron telescopes located on Mt. Sierra Negra in Mexico (at 4780 m) and Mt. Chacaltaya in Bolivia (at 5,250 m) both observed a clear solar neutron signal [36] , which was also recorded by three different counters located in the Northern and Southern Hemispheres. This made it possible to compare the detection efficiency of a solar neutron telescope with that of a conventional neutron monitor. The detection efficiency ratios were found to be 1 and 0.7, for the neutron monitor and neutron telescope respectively, pertaining to the same area of both detectors.
Since the solar neutron telescope cuts low-energy neutrons of less than 30 or 40 MeV, its detection efficiency is also lower than that of the neutron monitor [37] . The neutron monitor is highly sensitive to neutrons with energy exceeding about 10 MeV [8] . It is worth noting here that the data suggests the involvement of neutrons produced by both the impulsive and gradual phases [36] .
The FERMI-LAT satellite also recently observed two gamma-ray events in association with M-class solar flares on March 7 and June 7, 2011 [38] . Again a long duration component lasting more than 14 hours was observed and the continuous emission of GeV gamma rays from the Sun was detected. This mechanism may indicate a different mechanism in the gamma-ray production process in addition to that responsible for the impulsive production of gamma rays, which is discussed in the final part of this paper. An effective summary on solar neutron research has been recently published in a book, which also contains more detailed bibliography [39] .
The aim of this paper is to present new results using the FIB detector on board the ISS.
Actually Chapter 2 introduces details of the new solar neutron telescope FIB detector,

2-1 SEDA-AP-FIB detector
The new solar neutron telescope has been designed as a component of SEDA-AP. A detector for Space Environment Data Acquisition equipment -Attached Payload (SEDA-AP) was originally proposed to measure radiation levels at the International Space Station (ISS) in 1991 [12][13] .
In 2001, an actual Flight Module (FM) was ready to be deployed, but an accident involving the Space Shuttle resulted in the FM being stored in a special clean room for eight years until it could finally be launched.
SEDA-AP was designed as one of the detectors on board the Japan Exposure Module (JEM). This equipment not only comprises a neutron detector but also various other detectors, such as charged particle detectors, a plasma detector, an atomic oxygen monitor, and electronic device evaluation equipment. The system even includes a micro-particle capture detector. would be highly desirable to extend this period to cover at least one solar cycle of 11 years, provided that the system continues to operate.

2-2 The experimental purposes
This experiment has three main scientific goals as follows: (1) Accurate measurements of radiation levels in the ISS environment [40][41] (2) Rapid prediction of the imminent arrival of numerous charged particles from the Sun by monitoring GeV GLE particles for the flares of the western part of the solar surface (space weather forecast). However for the flares of the eastern part of the solar surface, the amount of emitted high energy particles may be estimated by observing neutrons.
(3) Identification of the production time of neutrons induced by the accelerated protons above the solar surface. We wish to know when and how high-energy particles are produced over the solar surface. When high-energy charged particles arrive at Earth and are detected, important information may be lost concerning the production time at the Sun. To understand the acceleration mechanism of charged particles at the Sun, it is necessary to compare the data of neutrons and gamma rays with images taken by a soft X-ray telescope [42] , RHESSI and/or the UV telescope launched on the Solar Dynamical To determine the neutron production time at the Sun, it is necessary to employ a neutron detector capable of measuring the energy of neutrons.
Currently, no such detector has been used in space other than an FIB detector, although the ground-based Solar Neutron Telescopes (SONTEL) have been operating for a number of years. [44][45][46] Accordingly, the FIB detector installed in SEDA-AP may provide a crucial data measuring neutron energy in space in the solar cycle 24.

2-3 Sensor design, detection efficiency, and trigger
To achieve the scientific goals listed earlier, a fine-grated neutron detector FIB has been designed, consisting of a plastic scintillator with 32 layers (sheets) and dimensions of 3 mm It can also identify the direction of neutron incidence. Neutrons and protons are discriminated by an anti-coincidence system consisting of six scintillator plates surrounding the FIB sensor in a cubic arrangement. To measure the total radiation dose at the ISS, we actually collect neutron data obtained over the South Atlantic Anomaly (SAA) region. The maximum count rate of the anti-coincidence system for the SAA region is 60,000 counts per second, and it works.
The cubic-shaped sensor used for neutron detection has sides measuring 10 cm and maximum kinetic energy of about 120 MeV. As shown in Figure 3, the sensor is monitored from two directions by two multi-anode photomultipliers (PMT1 and PMT2), meaning the arrival direction of the tracks can be identified. To determine the arrival direction of neutrons, protons must penetrate at least four sensor layers, each of which consists of plastic bars 3-mm thick. Consequently, the lowest neutron energy that can be measured is 35 MeV.
A trigger signal is produced by dynode signals from the PMTs (it is set at ≳ 30MeV proton equivalent). When the dynode signals from both PMTs exceed a certain threshold, a trigger signal is produced. When the trigger rate is less than 2 counts/sec, all ADC values for each channel are recorded in memory. The analog memory can handle all 512 channels of both PMTs. When the trigger rate exceeds 2 counts/sec, only the on-off signal (1 or 0) of each channel is recorded. When it exceeds 15 counts/sec, only the total output signal of the dynode is recorded. The technical details can be found elsewhere. [48][49][50]      As evidence that the FIB detector has been working stably, the energy spectrum of neutron-converted-protons is given in Figure 7.

Search for solar neutrons by the FIB detector
According to the calculations of Imaida [47] and Watanabe [51] , the typical event rate induced by solar neutrons is expected to be within the range 10 to 1,000 counts/sec in FIB.
As mentioned earlier, the background rate is as low as 0.

List of flares and search conditions
We searched solar neutrons for every solar flare of GOES X-ray intensity exceeding M2, the results of which are summarized in Table I

Background
Before introducing the actual neutron events, let us briefly describe the background.
As shown in the third panel of Figure 6, when the ISS approaches the northern or southern polar regions from the Equator, the neutron counting rate of the FIB detector increases from 0.04 to 0.5 counts/sec. The ISS completes an orbit around Earth every 90 minutes.
Therefore, should solar neutrons arrive when the ISS passes over the Equator, high quality data are obtained. However, it also passes over either side of the polar regions during each 45-minute period.
For this period, the FIB detector demonstrates new functions to discriminate background and squeeze signals. We assume that the direction of protons induced by solar neutrons was observed within a cone with an opening angle of about 45 degrees relative to the direction of the Sun. By applying a simple acceptance calculation (1/2*π steradian / 4π steradian = 1/8), the background may be reduced to 1/8. We can therefore identify neutron signals under a background level of 0.028 counts/sec (= 0.22/8). In other words, when the intensity of solar neutrons is weaker than 0.028 counts/sec, solar neutrons will become far more difficult to detect.

Actual event observed on March 7, 2011
On March 7, 2011, in association with the M3.7 flare, a possible signal of solar neutrons was captured by the FIB detector and more than 54 proton events were identified as coming from the direction of the Sun. In Figure 9, we present a distribution of the arrival direction of those 54 events over the background at the same time.
The statistical significance of the event was 6.8σ (based on the Li-Ma method). We regard those protons as being produced by solar neutrons inside the FIB sensor. Figure 10 shows     The FIB sensor has a function of measuring the energy of neutron-induced-protons.
Therefore the flight time of neutrons can be estimated. The result is shown in Figure 12.
In making Figure 12, the events are used that were emitted in the forward cone with an opening angle of less than 20 degrees. Figure 12 suggests that neutrons were emitted from the Sun during 19:41 and 19:54UT. It is worthwhile to note that the highest channel

An interpretation on the Time of Flight distribution of neutrons
The neutron-induced-protons of Figure  Taking account of these facts; (1) the angular distribution presented in Figure 9, (2) the production time distribution of neutrons (presented in Figure 12)  A sharp peak is recognized at the solar direction, while beyond 10 degrees, almost flat distribution is seen.

Other events
Three months later, the FIB detector observed another neutron event. This time the flare's intensity was only M2.5, hence far below the X-class scale. A total of 36 neutron events were identified. The statistical significance of the event was 5.8σ. The enhancement was observed during 06:21 and 06:41UT. Figure 14 shows the angular distribution of those events from the solar direction. On this event, the direction of the Sun was the opposite side of the pressurized module of ISS. So the identification from the background was easily made. The discrimination was also made by confirming the Bragg peak.

Discussions and Comparisons with Other Observations
A solar neutron event was observed on March 7, 2011, followed by another on June 7, 2011. Both provided a new perspective regarding the production process of solar neutrons.
To date, neutrons have been observed on the occasion of strong solar flares with X-ray intensity of ≳ X10 [52] . However, the present results indicate that even solar neutrons are  [53] respectively. Moreover, neutron monitors or solar neutron telescopes observed no solar neutrons from August 2009 to the end of July 2012, which may suggest that the soft X-ray flux measured by the GOES satellite does not necessarily correspond to the intensity of solar neutrons from the Sun. Although the link between this fact and the magnetic field structure [54] remains unclear, it remains a fascinating subject to be studied.
Another surprising fact is for such medium-class solar flares as those that occurred on March 7 (M3.7) and June 7, 2011 (M2.5), the LAT detector on board the Fermi satellite had observed long-lasting gamma-ray emissions. The emission of gamma rays with energies of 100 MeV to 1 GeV continued for more than 14 hours [38] . This marked a new discovery as the EGRET detector on board the CGRO satellite observed such long-lasting gamma-ray emissions with energy exceeding 150 MeV for the flare event on June 11, 1991 [35] . The RHESSI satellite observed the flare event on March 7 from its start time to the peak time, although the time profile of hard X-rays showed no unusual features and had a normal shape. The telescope of the Solar Dynamical Observatory (SDO) observed a very interesting feature; Coronal Mass Ejection (CME) started before the flare observed by the GOES/RHESSI satellites. A prominent injection of hot plasma into the base of the CME via loops was also observed by the SDO telescope. The signal of neutrons detected by the SEDA-FIB might be produced after that at the top of the inverse U-shaped loop around 19:58 UT [28] , while the long-lasting high-energy gamma rays of FERMI-LAT may be produced by the precipitation of the accelerated protons on the solar surface. The activity continued for more than 10 hours [38] .
For the solar flare event on June 7, 2011, hard X-ray data are available from the Fermi-GBM and RHESSI detectors. Once again, the telescope of the Solar Dynamical Observatory took a very interesting picture of this flare. The UV detector (sensitive to 171 nm) observed the very large-scale precipitation of plasma onto the solar surface [43] . In coincidence with this precipitation, protons trapped inside the upper "plasma bag" may precipitate over the solar surface. What resembles a "blow brush" over the reconnection point emits the high-energy protons confined in the plasma bag onto the solar surface, at which time the long-lasting gamma rays may be produced. However, the neutrons observed by the FIB detector may be produced when the protons accelerate, when magnetic loops are reconnected on top of the solar surface rather impulsively.
The UV telescope of SDO, however, did not observe any precipitation of plasma for the flare event on March 7, 2011, but instead a clear picture of the beginning of Coronal Mass Ejection (CME). After the seed of CME had been ejected, all loops under the CME started shining brightly. The protons were probably accelerated and neutrons possibly produced in one such loop via loop-to-loop interactions [55] when those protons struck the solar surface.  Table III Table I correspond to the event date, maximum time of X-ray intensity, and flare size respectively. The fourth column indicates the ISS location regarding whether on the shadow or sunny side of Earth (X or ○). The fifth column indicates whether solar neutrons are involved in the data. The ○→ X and X→○ notations in the fourth column indicate that the ISS was moving from the sunny side to the shadow side or vice versa 30 minutes from the peak flare time. The ? mark of the fifth column indicates a possible neutron event with statistical significance of less than 3σ . third, represent protons induced by neutrons, the incidents of which are consistent with those coming from the Sun, low-energy events difficult to identify as solar neutrons and ambiguous events due to low energy respectively. The marks ○, Δ, and × imply that each satellite was passing over the day side (○), partial day side (Δ) or night side (×) of Earth respectively. The event observed on September 8, 2011 was not involved in this Table, because it was observed near the SAA.  Table III. A comparison of solar neutron events observed by the SEDA-FIB detector with SMM event. The average trigger rate is simply obtained by dividing the total number of events by the observation time, while the mean flux is obtained taking the detection efficiency of both detectors into account (0.021 and 0.3 for SEDA and SMM) and also the detector area (100 and 450 cm 2 for SEDA and SMM).