Content uploaded by Martin J. Losekamm
Author content
All content in this area was uploaded by Martin J. Losekamm on Mar 23, 2016
Content may be subject to copyright.
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for
Space Applications
Thomas Pöschl∗
Technische Universität München
Physik-Department E18, James-Franck-Str 1, 85748 Garching, Germany
E-mail: thomas.poeschl@ph.tum.de
Martin J. Losekamm
Technische Universität München
Physik-Department E18, James-Franck-Str 1, 85748 Garching, Germany
Institute of Astronautics, Boltzmannstr. 15, 85748 Garching, Germany
E-mail: m.losekamm@tum.de
Daniel Greenwald
Technische Universität München
Physik-Department E18, James-Franck-Str 1, 85748 Garching, Germany
Stephan Paul
Technische Universität München
Physik-Department E18, James-Franck-Str 1, 85748 Garching, Germany
Cosmic antimatter is an excellent probe for many astrophysical processes. The abundancies and
energy spectra of antiparticles help to understand the creation and propagation mechanisms of
cosmic ray particles in the universe. The flux of geomagnetically trapped antiprotons in Earth’s
inner radiation belt can help to verify our models of the production of secondary cosmic ray parti-
cles and their motion in Earth’s magnetic field. The Antiproton Flux in Space (AFIS) experiment
will measure this flux using a novel CubeSat-sized particle detector. This active-target detector
consists of 900 scintillating fibers read out by silicon photomultipliers and is sensitive to antipro-
tons in the energy range below 100 MeV. It has a large geometrical acceptance of about 250cm2sr
because of its isotropic acceptance for particles. The particle identification scheme for antiprotons
relies on a combination of Bragg curve spectroscopy and the characteristics of the annihilation
process. In order to verify the detection principle, a prototype detector with a reduced number of
channels was tested at a stationary proton beam. Its energy resolution was found to be less than
1 MeV for stopping protons of about 50 MeV energy. We give an overview of the AFIS mission
and explain the working principle of the detector. We also discuss the results from the beam test.
The 34th International Cosmic Ray Conference,
30 July- 6 August, 2015
The Hague, The Netherlands
∗Speaker.
c
Copyright owned by the author(s) under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike Licence. http://pos.sissa.it/
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
1. Introduction
Particles created in Earth’s atmosphere can be trapped in belt-like structures around Earth by
the terrestrial magnetic field. These so-called Van Allen belts—named after their discoverer James
A. Van Allen—consist mainly of electrons and protons and were first detected in 1958 [18, 19].
Many have predicted the existence of a similar structure for antiprotons [6, 14, 8]. Due to the tilt and
shift of Earth’s magnetic dipole field from its rotational axis, the terrestrial magnetic field strength
is asymmetrically dependent on altitude around the planet [13] with an area of depression over the
southern Atlantic Ocean. In this so-called South Atlantic Anomaly (SAA), charged particles of the
inner Van Allen belt penetrate the atmosphere down to altitudes of 200km. This is the only area in
which a spacecraft in low-Earth orbit (LEO) can investigate this inner radiation belt.
The PAMELA experiment published the first measurements of the flux of trapped antipro-
tons in the SAA (Fig. 1) in 2011 [1]. Their measurements, based on 28 trapped antiprotons at
energies above approximately 80MeV, are two orders of magnitude smaller than theoretical pre-
dictions [14]. The AFIS experiment will complement this result by measuring the flux of trapped
antiprotons in the range of 25MeV to 100 MeV with a new active-target particle detector aboard a
nanosatellite in the SAA.
In this paper we report the status of the experiment: section 2describes the AFIS detector appa-
ratus, and our expected signal for antiprotons; and section 3describes the response of a scaled-down
prototype to protons and an energy reconstruction scheme based on Bayesian inference techniques.
2. The Multi-purpose Active-target Particle Telescope
The centerpiece of the AFIS mission is the multi-purpose active-target particle telescope (MAPT).
10 100 1000
10- 3
10- 2
10- 1
100
101
102
103
Particle flux [GeV m2 s sr]-1
Kinetic energy [MeV]
p (x 10-6)
pbar Selesnick, 2007
SAA PAMELA
GCR P A M E L A
AFIS
Figure 1: Fluxes of cosmic ray and trapped antiprotons inside the SAA as measured by the
PAMELA experiment [1]; and theoretical predictions from [14]. The AFIS measurement region
(25MeV–100MeV) is highlighted.
2
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
Figure 2: Structural layout of the MAPT. The ADU is symmetrically surrounded by the front-end
electronics (S1E, S2E).
Its core, the so-called active-detection unit (ADU), is made of 900 scintillating plastic fibers ar-
ranged in 30 layers of 30 fibers each. Layers are rotated 90◦from their neighbors to allow 3D
tracking of charged particles. This setup allows an isotropic acceptance for particles resulting in a
large geometrical acceptance of about 250 cm2sr [11]. Each fiber is read out by a dedicated silicon
photomultiplier (SiPM) with a matched peak-sensitivity wavelength. We use a 3D-printed frame to
support the active core and mount it into the satellite structure that is built according to the CubeSat
standard [17] (see Fig. 2).
Processing of the SiPM signals and their digitization is accomplished by the front-end elec-
tronics embedded in the detector module. To accept the very high event rates inside the SAA, signal
processing of the individual channels is fully parallelized.
Twenty ASICs perform over-threshold analog-to-digital conversion of the signals [11]. Four
FPGAs perform the subsequent time-over-threshold measurement. We can measure the signals
of the SiPMs, which are 10 ns to 100ns long, with a resolution of about 500 ps. A fifth FPGA
combines the data and makes the trigger decision. The total power consumption of this setup is
less than 20 W, meeting the operational constraints aboard a nanosatellite.
2.1 Particle Identification
Unlike spectrometer-based detectors, the MAPT detector cannot directly determine a particle’s
rigidity or charge sign. Instead, we identify particle species by their stopping characteristics. We
cannot use this technique to separate particle from antiparticle; but we can identify antiparticles
by their characteristic annihilations: they create high-energy secondary particles originating from
a single vertex. We can clearly identify an antiproton if
•it stops in the detector, so we can measure its Bragg curve and distinguish it from a high-
energy particle;
•it annihilates inside the active detector volume; and
•its annihilation products include charged particles.
3
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
2.1.1 Bragg Curve Spectroscopy
A particle with energy below the minimal ionizing region (βγ ∼3) initially deposits a small
fraction of its energy per unit length in a medium. As its kinetic energy decreases, the stopping
power of the medium on it increases and the particle deposits an increasing fraction of its energy.
This increase proceeds more rapidly as the particle slows down, leading to the deposition of a
significant fraction of the particle’s energy at the end of its trajectory. The differential profile of
such energy deposition as a function of depth is called the Bragg curve [10]. It is a distinct function
of energy for each interacting particle species. Since the particle is completely stopped, the integral
of the Bragg curve is the initial kinetic energy of the particle. The depth at which the particle stops,
its range, depends on its charge number, atomic number, and initial velocity1. The maximum
energy loss per unit distance depends only on the total charge of the incoming particle [7]. These
dependencies can be exploited to identify the stopping particle. This technique is called Bragg
curve spectroscopy and is widely used in nuclear and medical physics [7].
The MAPT detector cannot directly measure the energy loss per unit length. Instead, it mea-
sures the integral of the Bragg curve discretized by the fibers along the particle’s track. Bragg curve
spectroscopy therefore requires accurate track fitting to extract the true track length per channel and
the total range of the particle inside the detector volume. Additionally, dead layers in the detector
volume lead to differences between the deposited and measured energies and must be accounted
for.
We can use this method to distinguish protons and antiprotons from other particles by their
stopping and reconstruct their initial kinetic energies. However, such Bragg curve spectroscopy
cannot distinguish between particle and antiparticle2. We will use characteristics of the annihilation
process of the antiproton to identify it.
For particles that do not completely stop in the detector, we can use an extension of this
technique for species identification based on the so-called ∆E–∆Emethod [16], which also cannot
be used to discern particle from antiparticle. It can be applied to particles that pass completely
through the detector without stopping, but also to antiprotons that annihilate in flight. The change
of the energy deposition from channel to channel along the particle’s track is used to extrapolate
the stopping curve and subsequently determine the particle’s characteristics. The extrapolation is
based on fitting to the data values predicted from the Bethe equation, stopping-power tables, or a
parameterization of the ∆E–∆Espace [16]. A first study of the capabilities of these methods using
test-beam data is discussed in section 3.
2.1.2 Antiproton Annihilation
The most obvious difference between protons and antiprotons is that the latter annihilate with
ordinary matter: the antiproton vanishes along with a nucleon of the surrounding matter, and the
annihilated mass (about 2 GeV) is converted into energy released in the form of new secondary
particles emerging from the reaction. If these particles are charged, we can see them as tracks in-
side the detector volume. Since the cross section for annihilation increases rapidly with decreasing
particle velocity, the annihilation process usually takes place at very-low kinetic energies. Most
1The dependency on the atomic number and initial energy arises from the energy dependence.
2Neglecting the Barkas Effect occurring at very low particle energies [2], which we cannot observe.
4
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
antiprotons will first slow down in the detector due to energy losses by ionization processes be-
fore annihilating into secondary particles emitted isotropically due to conservation of momentum.
In contrast, background arising from high-energy proton interactions create secondary particles
boosted into the initial direction of the proton. Approximately four detectable secondary particles
are created per annihilation event. Most are fast pions, but heavier nuclear fragments and nucleons
are also created via fragmentation of the matter nucleus3. While the pions are usually minimally
ionizing and escape the detector without depositing a significant fraction of their energy, the heavier
particles completely deposit their energies in the detector. This leads to a unique signature for the
antiproton in its total energy deposition and its energy distribution in the detector volume. These
characteristics are very rarely present in other high-energy reactions.
3. In-Beam Measurement of Low-Energy Protons
To evaluate the reconstruction of stopping particles and to investigate the performance of the
fiber–SiPM combination, we tested a scaled-down prototype detector in a low-energy proton beam
at the πM1 beam line of the Paul Scherrer Institute in Villigen, Switzerland, in late 2013. The
prototype has 128 channels arranged in 16 layers of 8 fibers each. Like in the MAPT, layers are
rotated 90◦from their neighbors. The central overlap region of the detector is 16mm by 16 mm
in cross section. We used scintillating plastic fibers of type SCSF-78 manufactured by Kuraray.
To increase the light yield and prevent optical cross talk, we wrapped them in aluminum foil and
mirrored them at one end. The SiPMs are glued to the fibers with optical glue on the unmirrored
end. We use PM3350 silicon photomultipliers manufactured by KETEK with an active area of
3 mm ×3mm and a typical breakdown voltage of about 27V. Signals were digitized by eight mez-
zanine sampling analog-to-digital converters (MSADCs) developed for the COMPASS experiment
at CERN [12] and triggered externally by an NaI scintillator in front of the detector.
3.1 Event Reconstruction
We calibrated the pulse height in each channel, and thus its gain, using a beam of minimum-
ionizing particles (MIPs)—a 375-MeV/cπ−beam. A MIP deposits approximately 0.4MeV in
each layer of the detector and traverses all layers. We reconstruct the tracks of the particles in
the detector using a track-finding algorithm based on a standard Hough transformation [9]. The
response of the detector to protons was evaluated by recording the signal–depth distributions at
four different beam momenta (see Fig. 3a).
Only about half of the kinetic energy of each stopping proton is measured in the detector. To
quantify the loss mechanisms and to reconstruct the initial energies of incoming protons, we de-
veloped a numerical model based on the range tables provided by the NIST Physical Measurement
Laboratory [3]. We took two important processes affecting the response of the detector into ac-
count: First, quenching of the organic scintillator at high ionization rates leads to a reduced light
yield. According to Birk’s empirical formula, the light yield depends on the ionization rate [4]:
dL
dx∝dE
dx
1+kB dE
dx,(3.1)
3The antiproton can annihilate with both free protons—hydrogen nuclei—and larger nuclei.
5
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
Depth [Fiber]
2 4 6 8 10 12 14 16
Measured energy [MeV]
1
1.5
2
2.5
3
3.5
4
4.5
5
300 MeV/c
325 MeV/c
350 MeV/c
375 MeV/c
(a) Distribution of the measured energy deposi-
tion for different proton beam momenta.
Depth [Fiber]
2 4 6 8 10
Energy [MeV]
1
2
3
4
5
6
7
8
9
10
11
measured energy (data)
measured energy (model)
deposited energy (model)
(b) Comparison of deposited and measured ener-
gies, for 325-MeV/c protons.
Figure 3: Response of the detector to stopping protons.
with hdL/dxithe light yield of the scintillator per unit length, hdE/dxithe energy loss of the travers-
ing charged particle per unit length, and kB Birks’ coefficient. The latter is a material-specific con-
stant and depends on the base material and the doping characteristics of the scintillator [15]. Since
it has not yet been measured experimentally for our material, we treated it as a free parameter in
our analysis.
The second loss mechanism is the saturation of the SiPMs at high light exposure, experienced
during low-energy proton events. It introduces an additional nonlinear behavior in the measurement
and was accounted for in the analysis as well.
3.2 Bayesian Analysis
We use the Bayesian Analysis Toolkit (BAT) software package [5] to compare the measured
data with the expectations from our numerical model using Bayesian inference techniques. The
posterior likelihood distributions of unknown parameters and their correlations can be evaluated
using Markov Chain Monte Carlo (MCMC) methods [5]. To calculate the probability distributions
for the free parameters, such as the initial energies of the particles and Birks’ coefficient, their val-
ues are varied and the logarithmic likelihood (log-likelihood) function for each set of parameters is
evaluated. We define the overall log-likelihood function as the sum of the log-likelihood functions
for all fiber layers:
lnL=∑
i
lnL(i)S(i)
meas|S(i)
pred,(3.2)
with S(i)
pred being the predicted signal and S(i)
meas the measured signal in the ith layer. The log-
likelihood function for a single fiber is calculated by evaluating the logarithm of the probability
of the data set given the prediction, which we calculate as the likelihood of the prediction given the
spread of the data. The measured distribution can be approximated by a Gaussian distribution with
median µand standard deviation σ. Both values are obtained by fitting the measured pulse height
6
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
spectra in each layer. Hence, we calculate
lnL(i)S(i)
meas|S(i)
pred=−1
2 x(i)−µ(i)
σ(i)!2
−ln√2πσ(i)+ln2
|{z}
for non−hit layers
(3.3)
with x(i)being the predicted median signal height in the ith fiber. Per definition, the signal value is
always positive. This formula is only valid if the measured distribution is far away from the origin.
In our case, this requirement is always fulfilled except for layers that were not hit. In the calculation
of the log-likelihood function for these layers, an additional factor of ln2 must be added.
3.3 Results
Using the MCMC method, we obtained the posterior probability distributions of the unknown
parameters and the most probable set of parameters by maximizing the log-likelihood function.
This set describes the measured data well, as shown in Fig. 3b. In addition, the energy de-
posited in the fiber can be calculated. The combined fit of all proton data results in a value of
(0.127±0.030)mm/MeV for the Birks’ coefficient of our scintillating fibers. We used this method
to reconstruct the initial energies of particle beams with precision in the sub-MeV range for the en-
ergy range of 37MeV to 63MeV. We take this as an upper limit on the energy resolution for
protons of the MAPT detector.
4. Summary
The AFIS mission will measure the flux of trapped low-energy antiprotons in the inner Van
Allen belt. To fulfill all the requirements for applications on nanosatellites, we developed an active-
target particle detector comprised of scintillating plastic fibers and silicon photomultipliers. The
detector is sensitive to antiproton energies below 100 MeV and has a large geometrical acceptance
of about 250cm2sr. Hence, it will be possible to collect a statistically significant dataset to measure
the flux of low-energy antiprotons in the SAA. Antiproton identification relies on a combination of
Bragg-curve spectroscopy and the identification of the characteristics of antiproton annihilation.
A scaled-down prototype was built to investigate the response of the scintillating fibers and
SiPMs to ionizing radiation and stopping protons. We tested it in a particle beam and concluded
that it is possible to detect both minimum-ionizing radiation and highly ionizing radiation. Anal-
ysis of the data revealed two major processes affecting the reconstruction of energy deposited in
the detector volume: ionization quenching of the plastic scintillators and saturation effects in the
SiPMs. By comparing numerical simulations to experimental data, we quantified these effects and
reconstructed the initial energies of the stopped protons. We determined the initial energies of the
protons with an uncertainty of less than 1MeV in the energy range of 37MeV to 63MeV. This
value can be used as an estimate of the upper limit of the resolution of the detector for protons. The
analysis method based on Bayesian inference techniques will be extended to single event analysis
and to provide particle identification for the AFIS mission.
7
PoS(ICRC2015)590
A Novel CubeSat-Sized Antiproton Detector for Space Applications Thomas Pöschl
5. Acknowledgments
This research was supported by the DFG Cluster of Excellence Origin and Structure of the
Universe.
References
[1] Adriani, O., Barbarino, G. C., Bazilevskaya, G. a., et al. (2011). The Discovery of Geomag-
netically Trapped Cosmic-Ray Antiprotons. The Astrophysical Journal,737(2):L29.
[2] Barkas, W. H., Dyer, J. N., & Heckmann, H. H. (1963). Resolution of the ε−-Mass Anomaly.
Physical Review Letters,11(1):26–28.
[3] Berger, M. J., Coursey, J. S., & Zucker, M. A. (1999). ESTAR, PSTAR, and ASTAR: Computer
Programs for Calculating Stopping-Power and Range Tables for Electrons, Protons, and Helium
Ions (version 1.21).
[4] Birks, J. B. (1964). The Theory and Practice of Scintillation Counting. Pergamon Press.
[5] Caldwell, A., Kollár, D., & Kröninger, K. (2009). BAT â ˘
A¸S The Bayesian analysis toolkit.
Computer Physics Communications,180(11):2197–2209.
[6] Fuki, M. (2005). Cosmic-Ray Antiproton Spatial Distributions Computed In Magnetosphere.
International Journal of Modern Physics A,20(29):6739–6741.
[7] Gruhn, C., Binimi, M., Legrain, R., et al. (1982). Bragg curve spectroscopy.
[8] Gusev, A., Pugacheva, G., Pankov, V., et al. (2008). Antiparticle content in the magnetosphere.
Advances in Space Research,42(9):1550–1555.
[9] Illingworth, J. & Kittler, J. (1988). A Survey of the Hough Transform. Computer Vision,
Graphics, and Image Processing,44(1):87–116.
[10] Knoll, G. F. (1979). Radiation Detection and Measurement. Wiley.
[11] Losekamm, M. J., Pöschl, T., Langer, M., & Paul, S. (2014). The AFIS Detector: Measuring
Antimatter Fluxes on Nanosatellites. In 65th International Astronautical Congress.
[12] Mann, A. B., Konorov, I., Angerer, H., et al. (2009). The universal sampling ADC readout
system of the COMPASS experiment. In IEEE Nuclear Science Symposium Conference Record,
pages 2225–2228.
[13] Maus, S., Macmillan, S., McLean, S., et al. (2010). The US/UK World Magnetic Model for
2010-2015. NOAA Technical Report NESDIS/NGDC.
[14] Selesnick, R. S., Looper, M. D., Mewaldt, R. A., & Labrador, A. W. (2007). Geomagnetically
trapped antiprotons. Geophysical Research Letters,34(20):2–5.
[15] Tadday, K. A. (2011). Scintillation Light Detection and Application of Silicon Phtomulti-
pliers in Imaging Calorimetry and Positron Emission Tomography. PhD thesis, University of
Heidelberg.
[16] Terasa, C., Labrenz, J., Kühl, P., et al. (2013). Extension of measurement capabilities of the
Electron Proton Helium INstrument aboard SOHO. In 33nd International Cosmic Ray Confer-
ence, number 1.
[17] The CubeSat Program (2014). CubeSat Design Specification. California Polytechnic State
University,Rev. 13.
[18] Van Allen, J. A. (1958). Observation of High Intensity Radiation by Satellites 1958 Alpha
and Gamma. Journal of Jet Propulsion,28(9):588–592.
[19] Van Allen, J. A. & Frank, L. A. (1959). Radiation Around the Earth to a Radial Distance of
107,400 km. Nature,183(4659):430–434.
8
