Large-area balloon-borne polarized gamma ray observer (PoGO)

Abstract

We are developing a new balloon-borne instrument (PoGO), to measure polarization of soft gamma rays (25-200 keV) using asymmetry in azimuth angle distribution of Compton scattering. PoGO will detect 10% polarization in 100mCrab sources in a 6-8 hour observation and bring a new dimension to studies on gamma ray emission/transportation mechanism in pulsars, AGNs, black hole binaries, and neutron star surface. The concept is an adaptation to polarization measurements of well-type phoswich counter technology used in balloon-borne experiments (Welcome-1) and AstroE2 Hard X-ray Detector. PoGO consists of close-packed array of 397 hexagonal well-type phoswich counters. Each unit is composed of a long thin tube (well) of slow plastic scintillator, a solid rod of fast plastic scintillator, and a short BGO at the base. A photomultiplier coupled to the end of the BGO detects light from all 3 scintillators. The rods with decay times < 10 ns, are used as the active elements; while the wells and BGOs, with decay times ∼ 250 ns are used as active anti-coincidence. The fast and slow signals are separated out electronically. When gamma rays entering the field-of-view (fwhm ∼3deg²) strike a fast scintillator, some are Compton scattered. A fraction of the scattered photons are absorbed in another rod (or undergo a second scatter). A valid event requires one clean fast signal of pulse-height compatible with photo-absorption (> 20 keV) and one or more compatible with Compton scattering (< 10 keV). Studies based on EGS4 (with polarization features) and Geant4 predict excellent background rejection and high sensitivity.

Full text

Page 1

Large-Area Balloon-Borne Polarized Gamma Ray Observer (PoGO)
V. Andersson∗, P. Chen, T. Kamae, G. Madejski, T. Mizuno†, J.S.T. Ng, M. Suhonen∗, H. Tajima, T.
Thurston
SLAC and KIPAC, Stanford Univeersity, Menlo Park, California, USA
G. Bogaert
Ecole Polytechnique, Palaiseau, France
Y. Fukazawa
Hiroshima Univ., Higashi-Hiroshima, Japan
Y. Saito, T. Takahashi
ISAS/JAXA, Sagamihara, Japan
L. Barbier, P. Bloser, T. Cline, A. Harding, S. Hunter, J. Krizmanic, J. Mitchell, R. Streitmatter
NASA-GSFC, Greenbelt, Maryland, USA
R. Fernholz‡, E. Groth, D. Marlow
Princeton Univ., Princeton, New Jersey, USA
P. Carlson, W. Klamra, M. Pearce
Royal Inst. of Tech., Stockholm, Sweden
C.-I. Bjornsson, C. Fransson, S. Larsson, F. Ryde
Stockholm Univ., Stockholm, Sweden
M. Arimoto, T. Ikagawa, Y. Kanai, J. Kataoka, N. Kawai, Y. Yatsu
Tokyo Inst. of Tech., Tokyo, Japan
S. Gunji, H. Sakurai, Y. Yamashita
Yamagata Univ., Yamagata, Japan
We are developing a new balloon-borne instrument (PoGO), to measure polarization of soft gamma rays (30-
200 keV) using asymmetry in azimuth angle distribution of Compton scattering. PoGO is designed to detect
10 % polarization in 100mCrab sources in a 6-8 hour observation and bring a new dimension to studies on
gamma ray emission/transportation mechanism in pulsars, AGNs, black hole binaries, and neutron star surface.
The concept is an adaptation to polarization measurements of well-type phoswich counter consisting of a fast
plastic scintillator (the detection part), a slow plastic scintillator (the active collimator) and a BGO scintillator
(the bottom anti-counter). PoGO consists of close-packed array of 217 hexagonal well-type phoswich counters
and has a narrow field-of-view (∼ 5 deg2) to reduce possible source confusion. A prototype instrument has
been tested in the polarized soft gamma-ray beams at Advanced Photon Source (ANL) and at Photon Factory
(KEK). On the results, the polarization dependence of EGS4 has been validated and that of Geant4 has been
corrected.
1. Introduction
In the X-ray observation of extra-solar objects, po-
larization has been measured only once, by an ex-
ploratory and yet very successful experiment, of the
Crab (nebula+pulsar) at 2.6 and 5.2 keV with OSO-8
Satellite [1].Since then, polarization measurement
has been known to play crucial roles in high en-
ergy astrophysics.The next step towards improv-
ing our knowledge on celestial X-ray sources will re-
quire polarization measurements at the 10 % level
within characteristic time-scales of prominent flares
and transients (lasting less than a day) for flux level
of 50 − 100 mCrab.
Two detector technologies are now being developed
to reach the required sensitivity. The first technol-
∗Visitor from Royal Inst. of Technology
†Permanent address: Hiroshima University
‡Present address: NorthStar DesignWorks LLC, New Jersey,
USA
ogy, effective below ∼ 10 keV, is based on tracking of
the photoelectron from X-ray absorption in a micron-
scale gaseous imaging device. The other, effective be-
tween a few tens of keV to a few MeV, is based on the
coincidence measurement of Compton scattering and
photo-absorption. The direction of photoelectrons in
the first, and that of the scattered gamma rays in the
second, depend on the photon polarization.
PoGO is a novel instrument based on the Comp-
ton scattering technology: it is optimized in the hard
X-ray band (30-200 keV) and aims to resolve several
key physics issues via polarization measurements in a
series of balloon experiments lasting ∼6-8 hours each.
Compton scattering has highest potential for
measuring low polarization because the modulation
factor exceeds 85 % for a large solid angle between
60-120 degrees. Here the modulation factor is defined,
in term of the counting rate (R(φ)) at azimuthal
scattering angle φ, as:
(R(φ)max− R(φ)min)/(R(φ)max+ R(φ)min)
The other choice, the photoelectron tracking, operates
22nd Texas Symposium on Relativistic Astrophysics at Stanford University, Dec. 13-17, 2004
2508
1

Page 2

in the soft X-ray band and hence the instrument has
to be launched to a satellite orbit. It has, however,
the important advantage that a fine image can be ob-
tained by combining with a high throughput X-ray
mirror.
Science and instrumentation related to the polariza-
tion measurement of astronomical objects in the soft
and hard X-ray bands have been reviewed in articles
by Lei et al. [2] and by Dean [3].
A prototype PoGO detector with 7 units and the
flight-model components have been built and charac-
terized at synchrotron light facilities and with radio-
active sources. The two simulation programs added
with the polarization dependence, EGS4 [4] and
Geant4 [5], have been confronted with the experimen-
tal results and errors in Geant4 have been corrected.
The key features of the PoGO flight model given in
Table 1 have been calculated on the bug-fixed version
of Geant4.
2. Design of PoGO
The Well-type Phoswich Counter design [6] [7] has
been tested in several balloon experiments and proven
to be highly effective in reducing background [8]. The
technology has been subsequently adopted in the As-
troE/AstroE2 satellite mission as the Hard X-ray De-
tector [9] [10].The design presented here consists
of 217 well-type Phoswich Detector Cells (PDC, see
Fig.1) made of fast and slow scintillators, and a set
of anti-coincidence counters (Side Anti-Coincidence
Shield, SAS) made of BGO. The 217 PDCs function
collectively as an active collimator, an active shield,
and a Compton polarimeter, as conceptually shown in
Fig.2.
There are three figures of merit in polarization mea-
surements. One is the modulation factor, the azimuth
angle response of the instrument for a 100 % polar-
ized source. The second is the signal to background
ratio: in real observation, non-aperture background
and source confusion will dilute the modulation fac-
tor significantly. The third figure of merit is the effec-
tive area usable for polarization measurement. These
three figures of merit are often coupled: the effective
Table I Key Features of PoGO Design with 217 Units
Energy range
Geometrical area
Effective area (40 − 50 keV)
Background (< 50 keV)
Expected mod. factor (30 − 80 keV)
Sensitivity (3σ) for
100 mCrab point sources
30-200 keV
934 cm2
230 cm2
∼ 10 mCrab
24 %
9.0 %
area and the modulation factor may have to be sacri-
ficed to improve the signal to background ratio.
The modulation factor of Compton scattering ex-
ceeds 85 % for scattering angle of 60-120 degrees. The
narrow field-of-view (∼ 5 deg2) defined by the long
slow scintillator hexagonal tube and the Cu/Sn/Pb
foils, as well as the hermetic BGO shield wall (SAS)
are essential in achieving the low background envi-
ronment of ∼ 10 mCrab between 30 and 50 keV (see
Fig.3). The third figure of merit is the effective area.
In the PoGO design, the modulation factor and the
effective area have to be balanced to give an opimum
design given in Table 1.
Among the four event types schematically shown in
Fig.2, the one labeled as “fully contained gamma ray”
is the event to be selected. The PDC produces pulses
with a fast decay time (∼2 nsec) from the fast plas-
tic scintillator rods, and/or, with a slow decay time
(∼200-300 nsec) from the well scintillators and/or
BGO. The SAS produces pulses with a slow decay
time from BGO. The valid signal is selected involving
the entire assembly and in a hierarchical sequence as
in AstroE-HXD [9] [10].
An Level-0 trigger (L0) will be intiated when the
anode output of a PDC is greater than 20 keV but
less than that expected for a charged cosmic ray cross-
ing. An Level-1 trigger (L1) will follow an L0 if the
Pulse Shape Discriminator (PSD) asserts that the hit
is a clean fast scintillator hit. The prototype PSD
consists of a pair of fast (integration time ∼ 50 ns)
and slow (integration time ∼ 2 µs shaping amplifiers.
Pulse wave forms (sampled at ∼ 10 MHz) from the
two shaping amplifiers, non-zero pulse-heights from all
shaping amplifier outputs and all hits in Side Anti-
concidence System will then be recorded.
ground, the fully contained events with 2 or more
clean fast scinitllator hits will be selected. The two
largest energy deposits will be regarded as the photo-
absorption site and the scattering site and consistency
with the Compton kinematics (Fig.4) will be checked.
On the
3. Tests with Prototype Detectors
Components of the PDC were characterized first
with radioactive γ-rays (60 keV from241Am) and then
an assembled prototype PDC was tested with elec-
trons from90Sr (representing charge cosmic ray back-
ground) as well as γ-rays from241Am (representing
source flux). We then put 2 prototype systems con-
sisting of 7 units of prototype PDC’s, Front-End Elec-
tronics, and Pulse Shape Discriminator (PSD) to po-
larized γ-ray beams at Argonne National Laboratory
(ANL) and High Energy Acclerator Research Organi-
zation (KEK).
The detector arrangements used in these test are
shown in Fig.6. Only the center unit was used in the
test with radioactive sources: the gamma-ray source
22nd Texas Symposium on Relativistic Astrophysics at Stanford University, Dec. 13-17, 2004
2508
2

Page 3

Hexagonal Slow Plastic Scintillator Tube
Fast Scintillator
BGO
3 cm
20 cm60 cm
Figure 1: One well-type Phoswich Detector Cell (PDC) consiting of a 60cm long active anti-coincidence hexagonal well
(2 mm thick) made of slow plastic scintillator (decay time: ∼200-300 ns), a 20 cm long detection part made of fast
plastic scintillator (decay time: ∼ 2 ns), and a ∼ 5 cm long BGO. All scintillators are optically coupled and read out by
one 1-inch PMT (Hamamatsu R7899EG). The slow scintillator well are wrapped with thin foils of Cu/Sn/Pb to
attenuate off-axis gamma-rays.
Fully contained
gamma-ray
Charged particle
induced gamma-ray
Partially contained
gamma-ray
Side entering
gamma-ray
Back entering
gamma-ray
passive
collimator
slow plastic
scintillator
fast plastic
scintillator
side BGO
bottom BGO
PMT
Figure 2: Conceptual drawing showing the assembly of
the Phoswich Detector Cells (PDCs) and the Side
Anti-coincidence Scintillators (SAS). The SAS surrounds
the tightly bound 217 PDCs. The lines represent cosmic
particle and gamma-ray paths and small circles the
energy depositions. The L0 and L1 event filterings (see
text) will reject all but the one labeled “The fully
contained hard X-ray.”
was placed in front of the PDC on the Z-axis while
the electron source laterally in the X-Y plane. In the
beam tests, the 7 units were arranged as shown by the
6 solid or dashed hexagons.
3.1. Tests on a Well-Unit with
Radioactive Sources
Various prototype detectors have been built, tested,
and characterized until now. Individual components
have been characterized with radioactive sources.
Some crucial results are:
• Transmission of scintillation light from the fur-
thest end of the slow hexagonal tube.
shows pulse heights at 6 locations along the tube
axis of β-ray crossing its side walls (2 × 2 mm):
Fig.5
energy (keV) energy (keV)
22
10 10
)
-1
keV
-2
cm
-1
Flux (c s
-8
10
-7
10
-6
10
-5
10
-4
10
Figure 3: Background predicted by computer simulation
(Geant4) due to primary cosmic rays and albedo (all
directions) gamma rays. The coincidence rate
(Edeposit> 3 keV) due to albedo photons (thick and thin
histograms with crosses) is smaller than that of a
10mCrab source (thick histograms) below 50keV. The
two background histograms are for Edeposit= 30 keV
(thick) and 100 keV (thin).
Incident Photon Energy [keV]
20200100
0
100
200
Final Energy [keV]
Scattered photon
Recoil electron
90 deg
120 deg
60 deg
90 deg
120 deg
Figure 4: Compton kinematics in the PoGO energy
range. The energy depositions at the scattering and
absorption sites are widely separated, facilitating
identification with modest energy resolution by the
plastic scintillator.
22nd Texas Symposium on Relativistic Astrophysics at Stanford University, Dec. 13-17, 2004
2508
3

Page 4

polarization
plane
30 degree
y
x
Beam Direction
26.4mm
49mm
27.8mm
30mm
Figure 5: Arrangements of the 7 units in the two beam
tests. The 7 solid hexagons show the arrangemnt used at
Advanced Photon Source (ANL) and the 6 dashed
hexagons and the hexagon at the center make the
arrangement used at Photon Factory (KEK)
the measured attenuation factor of 3.4 is ade-
quate to reject the charged cosmic-ray crossing
by the prototype PSD. We expect the attenua-
tion factor to improve significantly in the flight
model now being developed.
• Ability for the prototype Pulse Shape Discrim-
ination system to retrieve the undistorted soft
gamma-ray spectrum in the presence of ∼ 200−
300 Hz cosmic-rays (per PDC) has been demon-
strated in Fig.7 and 8.
prototype PSD separates clean fast scintilla-
tor hits (signal region) from background. The
60Am spectrum retrieved from the signal region
while90Sr was hitting the slow scintillator at
200−300 Hz background is compared with that
taken in background-free environment in Fig.8.
Our L0, L1, and subsequent software filtering
can reduce background by nearly 4 orders of
magnitude.
Fig.7 shows how the
• Polarization-dependent modulation has been
observed by a double scattering setup with
60Am.
3.2. Beam Test at ANL Advanced Photon
Sources
To validate the PoGO design concept and simula-
tion programs, we measured modulation factor with
a PoGO prototype at MUCAT 6ID-D station at Ad-
vanced Photon Source (APS) Argonne National Lab-
oratory in November 2003.
hexagonal array of 7 fast scintillators of dimension
The prototype was a
Ratio: 3.4
Pulse Height
Figure 6: Attenuation factors measured for five
prototype slow scintillator hexagonal tubes. The upper
panel shows pulseheight ditributions obtained with90Sr
crossing the tube vertically at 6 positions along the
60 cm tube. The lower panel shows the peak pulse height
as a function of the distance from the nearest end.
approximately of the flight PDC except for the di-
ameter of photo-multiplier tubes (PMT): 1.25 inch
for this prototype and 1 inch for the flight model.
The geometric arrangement of the 7 units is shown
by solid hexagons in Fig.6. The photon beam was 98–
99% plane-polarized and its energy was set to 60 keV,
73 keV, or 83 keV. The beam passed through the cen-
ter of the array along the z axis in Fig.6. The beam
intensity at the entrance to the experimental area was
∼ 107photons s−1: we attenuated it just in front of
the array so that the trigger rate will be less than a
few kHz. A separate paper [11] gives further details
on the experiment.
The array was rotated at 15 degree steps around its
center, or equivalently, the beam. Data were taken
in two modes: one requiring coincidence between the
center unit and any of the 6 peripheral units; the other
by single hits in the center unit. Distribution of the
energy deposition at the center unit and the total de-
tected energy is shown in Fig.9 for the 73 keV run.
The events in the area surrounded by dashed lines
satisfy the Compton kinematics: we see clear sep-
aration between the Compton events and the back-
ground. The azimuthal angle dependence of the num-
ber of events in the region is plotted for 3 possible co-
incidence combinations (the dotted curves in Fig.10).
The modulation curves were fitted to sinusoidal curves
22nd Texas Symposium on Relativistic Astrophysics at Stanford University, Dec. 13-17, 2004
2508
4

Page 5

Slow=0.3*Fast+10
Signal Region
Figure 7: Distribution of the fast (abscissa) and slow
(ordinate) shaper outputs for a mix of signal (represented
by241Am) and background (represented by90Sr) The
S/N ratio is roughly 1/20. Clean hits in the fast
scintillator will enter the signal region surrounded by
dashed lines.
Spctr after “Signal Region” cut
Spctr without Sr90 bkgnd
60keV
Spctr of the Signal
Figure 8: The pulse height distribution of the events
(60 keV) in the “Signal region” in Fig.7 (the blue
histogram) and that taken without background.
to obtain modulation factor of 0.421 ± 0.010. This
is consistent with the factor (0.423 ± 0.012) obtained
without coincidence requirement.
We obtained modulation factors of 0.402 ± 0.011
and 0.416 ± 0.010 for the 60 keV and 83 keV runs,
respectively.
3.3. Beam Tests at KEK Photon Factory
In December 2004, another beam test was con-
ducted at Photon Factory (KEK) with a 7 unit ar-
ray made of the flight model fast scintillators (Eljen
Technology) and the flight model PMT’s (Hamamatsu
R7899EG, 1 inch diameter). The geometrical arrange-
ment of scintillators is different from that used in the
energy deposition in the central scintillator (keV)
0 2040 6080100120
total energy deposition (keV)
0
20
40
60
80
100
120
Figure 9: Distribution of total energy (ordinate) and the
energy at the center unit (abscissa) for Ebeam= 73 keV.
Th region surrounded by dashed lines are consistent for
being Compton scattered fully-contained events.
Rotation Angle (degree)Rotation Angle (degree)
-150-150-100-100-50-50005050100100150150
normalized counts
0
500
1000
1500
2000
2500
3000
ch1 and 7
ch2 and 6
ch3 and 5
Figure 10: Modulation curve obtained for events in the
region surrounded by dashed lines in Fig.9. The best fits
to the data are shown by the dotted curves. The solid
curve is the prediction by our version of Geant4 (see
text).
test at Advanced Photon Source (ANL) as shown by
dashed hexagons in Fig.6
The energy coverage has been extended down to
30 keV in the KEK test. The azimuthal modullation
are shown in Fig.11 for Ebeam= 30 keV and in Fig.12
for Ebeam = 70 keV. A detailed analysis will be de-
scribed elsewhere. [12].
4. Design Optimization Based on
Calibrated Simulation Program
4.1. Validation of the polarization
dependent EGS4 and Geant4
We used EGS4 [4] and Geant4 [5] to study the in-
strument response to polarized gamma rays and back-
22nd Texas Symposium on Relativistic Astrophysics at Stanford University, Dec. 13-17, 2004
2508
5