The Astrophysical Journal, 799:86 (24pp), 2015 January 20 doi:10.1088/0004-637X/799/1/86
C
2015. The American Astronomical Society. All rights reserved.
THE SPECTRUM OF ISOTROPIC DIFFUSE GAMMA-RAY EMISSION BETWEEN 100 MeV AND 820 GeV
M. Ackermann
1
, M. Ajello
2
, A. Albert
3
, W. B. Atwood
4
, L. Baldini
5
, J. Ballet
6
, G. Barbiellini
7,8
, D. Bastieri
9,10
,
K. Bechtol
11
, R. Bellazzini
5
, E. Bissaldi
12
, R. D. Blandford
3
, E. D. Bloom
3
, E. Bottacini
3
, T. J. Brandt
13
, J. Bregeon
14
,
P. Bruel
15
, R. Buehler
1
,S.Buson
9,10
, G. A. Caliandro
3,16
, R. A. Cameron
3
, M. Caragiulo
17
, P. A. Caraveo
18
,
E. Cavazzuti
19
, C. Cecchi
20,21
, E. Charles
3
, A. Chekhtman
22,67
, J. Chiang
3
, G. Chiaro
10
, S. Ciprini
19,23
, R. Claus
3
,
J. Cohen-Tanugi
14
, J. Conrad
24,25,26,68
, A. Cuoco
25,27,28
, S. Cutini
19,23
, F. D’Ammando
29,30
, A. de Angelis
31
,
F. de Palma
17,32
, C. D. Dermer
33
,S.W.Digel
3
, E. do Couto e Silva
3
, P. S. Drell
3
, C. Favuzzi
17,34
, E. C. Ferrara
13
,
W. B. Focke
3
, A. Franckowiak
3
, Y. Fukazawa
35
,S.Funk
3
,P.Fusco
17,34
, F. Gargano
17
, D. Gasparrini
19,23
,
S. Germani
20,21
, N. Giglietto
17,34
,P.Giommi
19
, F. Giordano
17,34
, M. Giroletti
29
, G. Godfrey
3
, G. A. Gomez-Vargas
36,37
,
I. A. Grenier
6
,S.Guiriec
13,69
, M. Gustafsson
38
, D. Hadasch
39
, K. Hayashi
40
, E. Hays
13
,J.W.Hewitt
41,42
, P. Ippoliti
29
,
T. Jogler
3
,G.J
´
ohannesson
43
,A.S.Johnson
3
,W.N.Johnson
33
, T. Kamae
3
, J. Kataoka
44
,J.Kn
¨
odlseder
45,46
,M.Kuss
5
,
S. Larsson
24,25,47
, L. Latronico
27
,J.Li
48
,L.Li
25,49
,F.Longo
7,8
, F. Loparco
17,34
,B.Lott
50
, M. N. Lovellette
33
,
P. Lubrano
20,21
,G.M.Madejski
3
, A. Manfreda
5
, F. Massaro
51
, M. Mayer
1
, M. N. Mazziotta
17
,J.E.McEnery
13,52
,
P. F. Michelson
3
, W. Mitthumsiri
53
, T. Mizuno
54
, A. A. Moiseev
42,52
, M. E. Monzani
3
, A. Morselli
36
, I. V. Moskalenko
3
,
S. Murgia
55
, R. Nemmen
13,41,42
,E.Nuss
14
, T. Ohsugi
54
, N. Omodei
3
, E. Orlando
3
,J.F.Ormes
56
, D. Paneque
3,57
,
J. H. Panetta
3
, J. S. Perkins
13
, M. Pesce-Rollins
5
, F. Piron
14
, G. Pivato
5
, T. A. Porter
3
,S.Rain
`
o
17,34
,R.Rando
9,10
,
M. Razzano
5,70
, S. Razzaque
58
,A.Reimer
3,39
,O.Reimer
3,39
, T. Reposeur
50
,S.Ritz
4
,R.W.Romani
3
,M.S
´
anchez-Conde
3
,
M. Schaal
59,67
, A. Schulz
1
, C. Sgr
`
o
5
,E.J.Siskind
60
, G. Spandre
5
, P. Spinelli
17,34
, A. W. Strong
61
,D.J.Suson
62
,
H. Takahashi
35
, J. G. Thayer
3
, J. B. Thayer
3
, L. Tibaldo
3
, M. Tinivella
5
, D. F. Torres
48,63
, G. Tosti
20,21
, E. Troja
13,52
,
Y. Uchiyama
64
, G. Vianello
3
, M. Werner
39
,B.L.Winer
65
,K.S.Wood
33
, M. Wood
3
, G. Zaharijas
12,66
, and S. Zimmer
24,25
1
Deutsches Elektronen Synchrotron DESY, D-15738 Zeuthen, Germany; markus.ackermann@desy.de
2
Department of Physics and Astronomy, Clemson University, Kinard Lab of Physics, Clemson, SC 29634-0978, USA
3
W. W. Hansen Experimental Physics Laboratory, Kavli Institute for Particle Astrophysics and Cosmology, Department of Physics and
SLAC National Accelerator Laboratory, Stanford University, Stanford, CA 94305, USA
4
Santa Cruz Institute for Particle Physics, Department of Physics and Department of Astronomy and Astrophysics, University of
California at Santa Cruz, Santa Cruz, CA 95064, USA
5
Istituto Nazionale di Fisica Nucleare, Sezione di Pisa, I-56127 Pisa, Italy
6
Laboratoire AIM, CEA-IRFU/CNRS/Universit
´
e Paris Diderot, Service d’Astrophysique, CEA Saclay, F-91191 Gif sur Yvette, France
7
Istituto Nazionale di Fisica Nucleare, Sezione di Trieste, I-34127 Trieste, Italy
8
Dipartimento di Fisica, Universit
`
a di Trieste, I-34127 Trieste, Italy
9
Istituto Nazionale di Fisica Nucleare, Sezione di Padova, I-35131 Padova, Italy
10
Dipartimento di Fisica e Astronomia “G. Galilei,” Universit
`
a di Padova, I-35131 Padova, Italy
11
Kavli Institute for Cosmological Physics, University of Chicago, Chicago, IL 60637, USA; bechtol@kicp.uchicago.edu
12
Istituto Nazionale di Fisica Nucleare, Sezione di Trieste, and Universit
`
a di Trieste, I-34127 Trieste, Italy
13
NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA
14
Laboratoire Univers et Particules de Montpellier, Universit
´
e Montpellier 2, CNRS/IN2P3, F-34095 Montpellier, France
15
Laboratoire Leprince-Ringuet,
´
Ecole Polytechnique, CNRS/IN2P3, F-91128 Palaiseau, France
16
Consorzio Interuniversitario per la Fisica Spaziale (CIFS), I-10133 Torino, Italy
17
Istituto Nazionale di Fisica Nucleare, Sezione di Bari, I-70126 Bari, Italy
18
INAF-Istituto di Astrofisica Spaziale e Fisica Cosmica, I-20133 Milano, Italy
19
Agenzia Spaziale Italiana (ASI) Science Data Center, I-00133 Roma, Italy
20
Istituto Nazionale di Fisica Nucleare, Sezione di Perugia, I-06123 Perugia, Italy
21
Dipartimento di Fisica, Universit
`
a degli Studi di Perugia, I-06123 Perugia, Italy
22
Center for Earth Observing and Space Research, College of Science, George Mason University, Fairfax, VA 22030, USA
23
INAF Osservatorio Astronomico di Roma, I-00040 Monte Porzio Catone (Roma), Italy
24
Department of Physics, Stockholm University, AlbaNova, SE-106 91 Stockholm, Sweden
25
The Oskar Klein Centre for Cosmoparticle Physics, AlbaNova, SE-106 91 Stockholm, Sweden
26
The Royal Swedish Academy of Sciences, Box 50005, SE-104 05 Stockholm, Sweden
27
Istituto Nazionale di Fisica Nucleare, Sezione di Torino, I-10125 Torino, Italy
28
Dipartimento di Fisica Generale “Amadeo Avogadro,” Universit
`
a degli Studi di Torino, I-10125 Torino, Italy
29
INAF Istituto di Radioastronomia, I-40129 Bologna, Italy
30
Dipartimento di Astronomia, Universit
`
a di Bologna, I-40127 Bologna, Italy
31
Dipartimento di Fisica, Universit
`
a di Udine and Istituto Nazionale di Fisica Nucleare, Sezione di Trieste, Gruppo Collegato di Udine, I-33100 Udine, Italy
32
Universit
`
a Telematica Pegaso, Piazza Trieste e Trento 48, I-80132 Napoli, Italy
33
Space Science Division, Naval Research Laboratory, Washington, DC 20375-5352, USA
34
Dipartimento di Fisica “M. Merlin” dell’Universit
`
a e del Politecnico di Bari, I-70126 Bari, Italy
35
Department of Physical Sciences, Hiroshima University, Higashi-Hiroshima, Hiroshima 739-8526, Japan
36
Istituto Nazionale di Fisica Nucleare, Sezione di Roma “Tor Vergata,” I-00133 Roma, Italy
37
Departamento de Fis
´
ıca, Pontificia Universidad Cat
´
olica de Chile, Avenida Vicu
˜
na Mackenna 4860, Santiago, Chile
38
Service de Physique Theorique, Universite Libre de Bruxelles (ULB), Bld du Triomphe, CP225, B-1050 Brussels, Belgium
39
Institut f
¨
ur Astro- und Teilchenphysik and Institut f
¨
ur Theoretische Physik, Leopold-Franzens-Universit
¨
at Innsbruck, A-6020 Innsbruck, Austria
40
Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5210, Japan
41
Department of Physics and Center for Space Sciences and Technology, University of Maryland Baltimore County, Baltimore, MD 21250, USA
42
Center for Research and Exploration in Space Science and Technology (CRESST) and NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA
43
Science Institute, University of Iceland, IS-107 Reykjavik, Iceland
44
Research Institute for Science and Engineering, Waseda University, 3-4-1, Okubo, Shinjuku, Tokyo 169-8555, Japan
45
CNRS, IRAP, F-31028 Toulouse cedex 4, France
1
The Astrophysical Journal, 799:86 (24pp), 2015 January 20 Ackermann et al.
46
GAHEC, Universit
´
e de Toulouse, UPS-OMP, IRAP, F-31400 Toulouse, France
47
Department of Astronomy, Stockholm University, SE-106 91 Stockholm, Sweden
48
Institute of Space Sciences (IEEC-CSIC), Campus UAB, E-08193 Barcelona, Spain
49
Department of Physics, KTH Royal Institute of Technology, AlbaNova, SE-106 91 Stockholm, Sweden
50
Centre d’
´
Etudes Nucl
´
eaires de Bordeaux Gradignan, IN2P3/CNRS, Universit
´
e Bordeaux 1, BP120, F-33175 Gradignan Cedex, France
51
Department of Astronomy, Department of Physics and Yale Center for Astronomy and Astrophysics, Yale University, New Haven, CT 06520-8120, USA
52
Department of Physics and Department of Astronomy, University of Maryland, College Park, MD 20742, USA
53
Department of Physics, Faculty of Science, Mahidol University, Bangkok 10400, Thailand
54
Hiroshima Astrophysical Science Center, Hiroshima University, Higashi-Hiroshima, Hiroshima 739-8526, Japan
55
Center for Cosmology, Physics and Astronomy Department, University of California, Irvine, CA 92697-2575, USA
56
Department of Physics and Astronomy, University of Denver, Denver, CO 80208, USA
57
Max-Planck-Institut f
¨
ur Physik, D-80805 M
¨
unchen, Germany
58
Department of Physics, University of Johannesburg, P.O. Box 524, Auckland Park 2006, South Africa
59
National Research Council Research Associate, National Academy of Sciences, Washington, DC 20001, USA
60
NYCB Real-Time Computing Inc., Lattingtown, NY 11560-1025, USA
61
Max-Planck Institut f
¨
ur Extraterrestrische Physik, D-85748 Garching, Germany
62
Department of Chemistry and Physics, Purdue University Calumet, Hammond, IN 46323-2094, USA
63
Instituci
´
o Catalana de Recerca i Estudis Avan¸cats (ICREA), E-08028 Barcelona, Spain
64
3-34-1 Nishi-Ikebukuro, Toshima-ku, Tokyo 171-8501, Japan
65
Department of Physics, Center for Cosmology and Astro-Particle Physics, The Ohio State University, Columbus, OH 43210, USA
66
The Abdus Salam International Center for Theoretical Physics, Strada Costiera 11, Trieste I-34151, Italy
Received 2014 June 19; accepted 2014 September 29; published 2015 January 19
ABSTRACT
The γ -ray sky can be decomposed into individually detected sources, diffuse emission attributed to the interactions
of Galactic cosmic rays with gas and radiation fields, and a residual all-sky emission component commonly called
the isotropic diffuse γ -ray background (IGRB). The IGRB comprises all extragalactic emissions too faint or too
diffuse to be resolved in a given survey, as well as any residual Galactic foregrounds that are approximately
isotropic. The first IGRB measurement with the Large Area Telescope (LAT) on board the Fermi Gamma-ray
Space Telescope (Fermi) used 10 months of sky-survey data and considered an energy range between 200 MeV and
100 GeV. Improvements in event selection and characterization of cosmic-ray backgrounds, better understanding
of the diffuse Galactic emission (DGE), and a longer data accumulation of 50 months allow for a refinement
and extension of the IGRB measurement with the LAT, now covering the energy range from 100 MeV to
820 GeV. The IGRB spectrum shows a significant high-energy cutoff feature and can be well described over
nearly four decades in energy by a power law with exponential cutoff having a spectral index of 2.32 ± 0.02 and
a break energy of (279 ± 52) GeV using our baseline DGE model. The total intensity attributed to the IGRB is
(7.2 ±0.6) × 10
−6
cm
−2
s
−1
sr
−1
above 100 MeV, with an additional +15%/−30% systematic uncertainty due to
the Galactic diffuse foregrounds.
Key words: diffuse radiation – gamma rays: diffuse background
Supporting material: machine-readable table
1. INTRODUCTION
The universe is filled with electromagnetic radiation, which
can be characterized by a cosmological energy density and spec-
trum. This extragalactic background light (EBL) is energetically
dominated by thermal relic radiation from the last scattering sur-
face observed as the cosmic microwave background. Different
physical processes characterize the EBL in each waveband—
starlight in the optical, thermal dust emission in the infrared,
and emission from active galactic nuclei (AGNs) in X-rays. The
extragalactic γ -ray background (EGB) provides a nonthermal
perspective on the cosmos, which is also explored through the
cosmic radio background, as well as extragalactic cosmic rays
(CRs) and neutrinos.
The EGB represents a superposition of all γ -ray sources,
both individual and diffuse, from the edge of the Milky Way to
the edge of the observable universe, and is thus expected to en-
code diverse phenomena (see Dermer 2007 for a comprehensive
67
Resident at Naval Research Laboratory, Washington, DC 20375, USA.
68
Royal Swedish Academy of Sciences Research Fellow, funded by a grant
from the K. A. Wallenberg Foundation.
69
NASA Postdoctoral Program Fellow, USA.
70
Funded by contract FIRB-2012-RBFR12PM1F from the Italian Ministry of
Education, University and Research (MIUR).
review). Guaranteed contributions arise from established ex-
tragalactic γ -ray source classes including AGNs, star-forming
galaxies, and γ -ray bursts. The beamed emission from blazars
is sufficiently bright that statistically large samples of individ-
ual sources have now been detected to cosmological distances
(Ackermann et al. 2011a). Accordingly, the flux distribution
of blazars even below the detection threshold for individual
sources can in principle be estimated from a relatively firm
empirical basis through an extrapolation of the observed flux
distribution (e.g., Abdo et al. 2010c; Ajello et al. 2012, 2014),
although a consensus has not yet been reached (e.g., Singal
et al. 2012). For other populations, such as star-forming galax-
ies (Pavlidou & Fields 2002; Thompson et al. 2007; Fields et al.
2010; Makiya et al. 2011; Stecker & Venters 2011; Ackermann
et al. 2012c) and AGNs with jets oriented obliquely to our line
of sight (Inoue 2011;DiMauroetal.2014), the cumulative
intensity is almost entirely unresolved by current instruments;
calculations of the flux distribution incorporating physical mod-
els and/or multiwavelength scaling relations must be invoked
to estimate their EGB contributions. There are additional theo-
retically well-motivated extragalactic source classes too faint to
have been individually detected thus far, including galaxy clus-
ters and their associated large-scale structure formation shocks
(Colafrancesco & Blasi 1998; Loeb & Waxman 2000).
2
The Astrophysical Journal, 799:86 (24pp), 2015 January 20 Ackermann et al.
At energies 100 GeV, the interaction length for γ -rays
with photons of the UV/optical/IR EBL becomes much shorter
than a Hubble length, thus defining an effective γ -ray horizon.
The electromagnetic cascades initiated by both very high energy
γ -rays (Coppi & Aharonian 1997) and ultra high energy CRs
(Berezinskii & Smirnov 1975) interacting with the EBL create
truly diffuse EGB contributions.
Finally, more exotic processes such as dark matter
annihilation/decay may be present, though as yet unrecognized,
in the EGB (Bergstr
¨
om et al. 2001; Ullio et al. 2002; Taylor &
Silk 2003).
From an observational standpoint, there are two main chal-
lenges in measuring the EGB. One is to model the diffuse Galac-
tic emission (DGE) created by CR interactions with interstellar
gas (ISG) and interstellar radiation fields (ISRFs), which is com-
parable to the EGB intensity at energies 1 GeV even at the
Galactic poles and therefore represents a strong foreground to
the EGB measurement. The second challenge is separating cos-
mic γ -rays from CR-induced backgrounds at the detector level.
For instruments in low Earth orbit, the CR intensity can exceed
that of the EGB signal by a factor of up to ∼10
6
. In addition,
there is a sizable flux of secondary particles that are produced
by interactions of CRs in Earth’s atmosphere.
The existence of all-sky γ -ray emission was first realized
experimentally using 621 candidate γ -rays collected by the
OSO-3 satellite (Clark et al. 1968; Kraushaar et al. 1972),
while Fichtel et al. (1975, 1978) reported the first spectral
measurement of an isotropic diffuse background using the SAS-2
satellite. Analyses using more sensitive instruments capable of
detecting individual extragalactic sources began reporting the
residual all-sky average intensity after subtracting individual
sources and DGE templates (e.g., Sreekumar et al. 1998; Strong
et al. 2004, using EGRET). The remaining emission component
is found to be approximately isotropic on large angular scales
and is commonly called the isotropic diffuse γ -ray background
(IGRB). The sum of the IGRB and individually resolved
extragalactic sources represents an upper limit to the total EGB
intensity, since residual unresolved Galactic emissions may be
present in the IGRB. For example, CR interactions with gas
(Feldmann et al. 2013) or radiation fields (Keshet et al. 2004)in
the extended halo of the Milky Way, unresolved Galactic sources
such as millisecond pulsars (Faucher-Gigu
`
ere & Loeb 2010),
and CR interactions with solar system debris (Moskalenko &
Porter 2009) and the solar radiation field (Moskalenko et al.
2006; Orlando & Strong 2007, 2008) have been considered as
sources of approximately isotropic emission on large angular
scales.
The intensity attributed to the IGRB is observation dependent
because more sensitive instruments with deeper exposures
can extract fainter extragalactic sources, whereas the total
EGB intensity (assuming complete subtraction of all Galactic
emissions) is the fundamental quantity.
The Large Area Telescope (LAT) on board the Fermi Gamma-
ray Space Telescope (Fermi) is the first instrument with sufficient
collecting area and CR-background rejection power to measure
the IGRB at energies > 100 GeV. Since launch into low-
Earth orbit on 2008 June 11, the LAT has operated primarily
in a sky-survey mode that, combined with a large field of
view (2.4 sr) and good spatial resolution (∼1
◦
at 1 GeV), has
enabled the most detailed studies of the DGE to date. The
LAT is a pair-conversion telescope consisting of a precision
tracker and imaging calorimeter, which are used together to
reconstruct γ -ray directions and energies, and a surrounding
segmented anticoincidence detector (ACD) to identify charged
particles entering the instrument. Atwood et al. (2009) provide
a description of the Fermi mission and LAT detector, as well as
details of the on-orbit calibration. Ground data processing, event
selection, and instrument response functions (IRFs) are provided
in Abdo et al. (2009c), Ackermann et al. (2012b, 2012d).
A first measurement of the IGRB spectrum between 200 MeV
and 100 GeV based on 10 months of LAT data was published in
Abdo et al. (2010b). In this paper we present a refinement and
extension of that analysis based on 50 months of sky-survey
observations. Multiple improvements in event classification,
Galactic foreground and CR background models, and analysis
techniques have been implemented. Together with increased
statistics, these updates allow for an extension of the LAT IGRB
measurement by over a decade in energy, now covering the range
from 100 MeV to 820 GeV.
2. DATA SAMPLES
Fifty months of LAT data recorded between 2008 August 5
and 2012 October 6 are used for this analysis, corresponding
to a total observation time of 1239 days.
71
The events have
been reprocessed with an updated instrument calibration, which
improves the agreement between data and simulation of the en-
ergy reconstruction quality, the point-spread function (PSF), and
certain classification variables and thereby reduces systematic
uncertainties (Bregeon et al. 2013).
72
The LAT IGRB analysis poses especially stringent require-
ments on the γ -ray purity of the event selection since both
the signal and CR-background spatial distributions are quasi-
isotropic. The residual CR background contamination must be
reduced to a relatively small fraction of the total isotropic inten-
sity in order to measure the IGRB with acceptable systematic
uncertainty because the (not perfectly known) CR background
is directly subtracted from the total isotropic intensity in the
final step of evaluating the IGRB.
The predefined event classes publicly available from the
Fermi Science Support Center, including P7ULTRACLEAN,have
insufficient CR background rejection performance for the IGRB
analysis energies below E<400 MeV and energies above
E>100 GeV. Therefore, we developed two dedicated event
samples for the IGRB analysis with distinct selection criteria
at low and high energies. The IGRB intensity measurements
reported in Section 5 use the “low-energy” sample for the energy
range 100 MeV to 13 GeV and the “high-energy” sample for
the energy range 13–820 GeV.
73
Multiple event classifications
are necessary in order to obtain the best-possible compromise
between statistics and low CR backgrounds across the full LAT
energy range since the compositions and interactions of CR
71
LAT data recording is disabled for ≈13% of the total on-orbit time, during
passages through the South Atlantic Anomaly (SAA), a region with extremely
high charged particle backgrounds. Only observation periods that passed data
quality monitoring and where the angle between the LAT z-axis and the zenith
was below 52
◦
are used for this analysis. Note that the actual live time is ∼8%
smaller than the 1239 day observation time quoted here owing to instrumental
dead time associated with event latching and readout.
72
The reprocessed data are available from the Fermi Science Support Center
(http://fermi.gsfc.nasa.gov/ssc/), together with a list of caveats regarding their
usage.
73
Each sample includes events from the full LAT energy range. The labels
“low-energy” and “high-energy” refer to the energy regime for which the event
classifications have been optimized. The energy overlap between samples
allows for consistency checking between the low-energy and high-energy
analyses (described in Section 3), a feature we used to verify that the specific
choice of crossover energy between 5 and 50 GeV does not affect the accuracy
of our quoted results for the IGRB intensity.
3
The Astrophysical Journal, 799:86 (24pp), 2015 January 20 Ackermann et al.
Tab le 1
Event Selection Criteria for Low-energy and High-energy Samples, Including Modifications with Respect to P7ULTRACLEAN
Low-energy High-energy
Data Sample/Event Selection P7REP_IGRB_LO P7REP_IGRB_HI
Add tracker veto I Y Y
Add tracker veto II Y N
Add deposited charge veto Y N
Remove calorimeter shower maximum veto N Y
Incidence angle veto >72
◦
>72
◦
Zenith angle veto >90
◦
>105
◦
Notes. See Section 2.1 for detailed descriptions of the event selection criteria. The low-energy and high-energy
event samples are used to derive the IGRB intensity in the 100 MeV to 13 GeV and 13 GeV–820 GeV energy
ranges, respectively.
backgrounds in the low- and high-energy regimes are rather
different. The modifications to the baseline P7ULTRACLEAN
classification for the two event samples used in this work are
described below and summarized in Table 1.
2.1. Event Selection
The low-energy sample is a strict subset of events classified as
photons according to the P7ULTRACLEAN event class definition
(Ackermann et al. 2012b). To reduce the residual background
of secondary electrons, positrons, and protons produced by
CR interactions in the Earth’s atmosphere, which are the
primary concern in the low-energy IGRB analysis, the following
additional criteria are imposed.
Tracker veto I. Part of the tracker is used as an additional veto
to complement the ACD. Specifically, we require the re-
constructed γ -ray trajectory to cross at least two layers
of active silicon strip detector area without producing hits
in these detectors. This selection criterion significantly in-
creases the efficiency of vetoing charged particles entering
the LAT.
Tracker veto II. We discard events for which the reconstructed
pair-conversion vertex lies in the three x–y double layers of
the tracker closest to the calorimeter. Comparisons of low-
background and high-background on-orbit data sets, as well
as comparisons of γ -ray and CR-background Monte Carlo
simulations, have shown that these events suffer a higher
background contamination fraction.
Deposited charge veto. γ -rays convert into an electron–positron
pair, while most of the background events involve a
single charged particle. Therefore, we require the charge
deposited in the first tracker layer following the interaction
vertex to be >1.5 times the value expected for a minimum
ionizing particle, which is typically indicative that two
particles crossed the layer rather than one by itself.
Incidence angle veto. Events arriving from directions >72
◦
off
the LAT boresight are rejected because there is increased
CR background leakage for such highly inclined events.
The new event class for the low-energy sample is denoted
as P7REP_IGRB_LO in the remainder of this manuscript to
distinguish it from the publicly available standard event classes.
The sky-averaged exposure of the P7REP_IGRB_LO selection
is 66% of the exposure of the corresponding P7ULTRACLEAN
selection for survey mode observations (see Figure 1), when
compared at the energy of maximum exposure. The estimated
residual CR background rate is reduced by a factor of ∼3 around
200 MeV, where the background rate is highest.
Energy [MeV]
3
10
4
10
5
10
6
10
s]
2
Average exposure [cm
0
20
40
60
80
100
120
9
10
) < 90
zenith
P7REP_IGRB_LO (
) < 105
zenith
P7REP_IGRB_HI (
) < 90
zenith
P7REP_ULTRACLEAN_V15 (
Figure 1. Comparison of the sky-averaged exposure for the P7REP_IGRB_LO,
P7REP_IGRB_HI,andP7ULTRACLEAN event selections. Thick lines indicate the
respective energy ranges for which the P7REP_IGRB_LO and P7REP_IGRB_HI
event classes are used in this analysis.
As a final step to define the low-energy sample, events from
measured directions >90
◦
off the Earth zenith are rejected to
limit contamination by photons from the Earth limb (Abdo et al.
2009a).
For the high-energy sample, we use a relaxed event selection
compared to P7REP_IGRB_LO. At energies above 13 GeV, CR
primaries in the form of protons and heavier nuclei dominate
the background flux. The rejection power for CR nuclei is
sufficient for this analysis if one requires only the condition
described above as “Tracker veto I” in addition to the standard
P7ULTRACLEAN event class definitions. We implement the
“Incidence angle veto” as for the low-energy event class.
The standard P7ULTRACLEAN event classification rejects
events for which the positions of the primary interaction ver-
tex and the reconstructed shower maximum are separated by
>12 radiation lengths as measured along the shower axis. This
selection criterion was introduced to reject CR events with bad
shower reconstructions that would sometimes result in large
apparent depths for the shower maxima—a strategy that works
well for energies 500 GeV but removes a significant fraction
4
The Astrophysical Journal, 799:86 (24pp), 2015 January 20 Ackermann et al.
of γ rays 500 GeV. Therefore, this selection criterion is re-
moved in the event selection for the high-energy sample. The
very moderate increase in residual CR background arising from
this removal is overcompensated by the “Tracker veto I” condi-
tion that was introduced for this event class.
The distinct classification scheme for the high-energy
sample is denoted as P7REP_IGRB_HI in contrast to the
P7REP_IGRB_LO classification used for the low-energy sample.
At high energies, the reconstructed arrival directions of CR-
induced atmospheric γ rays are confined to angles very close to
the Earth limb (113
◦
from the Earth zenith) owing to the reduced
width of the PSF (∼0.
◦
1 above 10 GeV). Therefore, the zenith
angle veto condition described above for the low-energy sample
is modified to reject only photons from directions >105
◦
off the
Earth zenith.
The P7REP_IGRB_HI selection has a peak exposure of about
85% of the peak exposure of the standard P7ULTRACLEAN
selection, and it surpasses the P7ULTRACLEAN selection in
acceptance 700 GeV.
Both new event selections were cross-validated against the
standard P7ULTRACLEAN event selection by performing the
analysis described below also on the P7ULTRACLEAN data
set. We obtain consistent results in the energy range in which
we can perform the analysis even in the presence of the higher
CR background of the P7ULTRACLEAN selection (400 MeV
to 100 GeV).
2.2. Instrument Response Functions
New sets of dedicated IRFs were created for the low-energy
(P7REP_IGRB_LO) and the high-energy (P7REP_IGRB_HI)
event classes via Monte Carlo simulation of γ -rays. The en-
ergy range of the new IRFs is 17.8 MeV to 1.78 TeV.
Figure 1 shows the sky-averaged exposures obtained for the
low-energy and high-energy samples using the corresponding
P7REP_IGRB_LO and P7REP_IGRB_HI IRFs, respectively. The
exposure that would be obtained for the same observation
period, but using the standard P7_ULTRACLEAN event sample
with IRFs P7REP_ULTRACLEAN_V15, is plotted for comparison.
For the P7REP_IGRB_HI selection, there is an overall drop in
exposure as a result of using part of the silicon tracker to veto
charged particles. However, at the highest energies, especially
>300 GeV, this loss is increasingly counteracted by the removed
shower maximum constraint in the P7REP_IGRB_HI class (see
Section 2.1). The P7REP_IGRB_LO selection has a significantly
lower average exposure than P7REP_IGRB_HI. This loss of
exposure is acceptable at low energies where this event class
is used since the IGRB measurement is not limited by statistics
below tens of GeV.
In-flight PSF corrections available for the IRFs correspond-
ing to standard event classes have not been applied to the
P7REP_IGRB_LO and P7REP_IGRB_HI IRFs. The corrections
were motivated by small differences observed in the PSF of
the original (P7) on-orbit and simulated LAT data at energies
1 GeV (Ackermann et al. 2012b, 2013a). We verified directly
that such small corrections, mitigated in the reprocessed data
(Bregeon et al. 2013), do not significantly affect this analysis.
This is expected since it is performed on a spatial grid of about
0.
◦
9, considerably larger than the typical high-energy PSF.
2.3. Residual Cosmic-ray Background
Charged and neutral CRs misclassified as γ -rays by the mul-
tivariate event classification algorithms mimic an isotropic flux
that is indistinguishable from the IGRB. In addition, genuine
γ -rays from the Earth’s atmosphere that have directional recon-
struction errors sufficient to bypass the zenith angle veto become
a source of apparent extraterrestrial emission over the full sky.
In this work, the term “CR background” includes CR-induced
γ -rays from the atmosphere.
Our estimation of the residual CR background event rate is
based on Monte Carlo simulations of the relevant particle species
in the near-Earth environment, namely, CR nuclei and electrons,
as well as their atmospheric secondaries. We simulate both CR
backgrounds and signal γ -rays and extract the distributions for
reconstructed event properties with the greatest discrimination
power at low and high energies, respectively: the multivariate
event classifier output and the transverse shower size. The
distributions for simulated background and signal are compared
to the distributions for the flight data to quantify the level
and associated uncertainty of the CR background. A detailed
description of this method can be found in Ackermann et al.
(2012b).
To account for atmospheric γ -rays surviving the zenith
angle veto, an updated phenomenological model for the Earth
emission based on LAT observations is included in the Monte
Carlo simulation. Atmospheric γ -rays can bypass the zenith
veto either by being reconstructed in the extreme tail of the
PSF or by entering from the back side of the instrument
and being reconstructed as though coming from the front.
Although such catastrophic mis-reconstructions are rare, the
Earth emission is sufficiently bright that the expected event
rate is nonnegligible at energies 1 GeV (Bechtol 2012). For
the stringent zenith angle selections used in this work, the
residual contamination of atmospheric γ -rays is expected to be
composed primarily of back-entering events. The reconstructed
directional distribution of back-entering atmospheric γ -rays in
particular is approximately isotropic.
Figure 2 shows the residual CR background rates as a
function of reconstructed energy for the P7REP_IGRB_LO and
P7REP_IGRB_HI classes. Note that the event energy is recon-
structed under the hypothesis of a front-entering γ -ray and in
general does not represent the actual energy for hadrons. At high
energies, primary protons and electrons both contribute signif-
icantly to the CR background. Although protons are far more
abundant than electrons in the environment of the LAT, there is
also greater rejection power against protons since analysis of the
shower shape in the calorimeter can be used to tag and remove
protons in addition to the veto power obtained from the ACD.
All contributions shown in Figure 2 have been adjusted from the
raw Monte Carlo predictions based on event property compar-
isons between the simulated and flight data, as described above.
The CR background contamination after rescaling agrees with
the raw predictions from Monte Carlo simulation to within 35%,
depending on energy, and this maximum discrepancy is used as
a measure of the systematic uncertainty in the CR background
contamination.
The full uncertainty in residual CR background rates, shown
in Figure 2, has been calculated by adding systematic and
statistical uncertainties in quadrature. For the P7REP_IGRB_HI
event class at energies above 10 GeV (relevant for the high-
energy sample), the statistical uncertainties are large owing to
the limited size of the simulated residual background sample.
74
Therefore, instead of using bin-by-bin estimates for the CR
74
The existing background rate estimates were derived from several million
CPU hours of CR simulation. Significant gains in precision might be achieved
in the future when more computing power becomes available.
5
