Acceleration of Petaelectronvolt protons in the Galactic
Centre
H.E.S.S. Collaboration: A. Abramowski
1
, F. Aharonian
2,3,4
, F. Ait Benkhali
2
, A.G. Akhperjanian
5,4
,
E.O. Ang
¨
uner
6
, M. Backes
7
, A. Balzer
8
, Y. Becherini
9
, J. Becker Tjus
10
, D. Berge
11
, S. Bernhard
12
,
K. Bernl
¨
ohr
2
, E. Birsin
6
, R. Blackwell
13
, M. B
¨
ottcher
14
, C. Boisson
15
, J. Bolmont
16
, P. Bordas
2
,
J. Bregeon
17
, F. Brun
18
, P. Brun
18
, M. Bryan
8
, T. Bulik
19
, J. Carr
20
, S. Casanova
21,2
, N. Chakraborty
2
,
R. Chalme-Calvet
16
, R.C.G. Chaves
17,22
, A, Chen
23
, M. Chr
´
etien
16
, S. Colafrancesco
23
, G. Cologna
24
,
J. Conrad
25,26
, C. Couturier
16
, Y. Cui
27
, I.D. Davids
14,7
, B. Degrange
28
, C. Deil
2
, P. deWilt
13
,
A. Djannati-Ata
¨
ı
29
, W. Domainko
2
, A. Donath
2
, L.O’C. Drury
3
, G. Dubus
30
, K. Dutson
31
, J. Dyks
32
,
M. Dyrda
21
, T. Edwards
2
, K. Egberts
33
, P. Eger
2
, J.-P. Ernenwein
20
, P. Espigat
29
, C. Farnier
25
,
S. Fegan
28
, F. Feinstein
17
, M.V. Fernandes
1
, D. Fernandez
17
, A. Fiasson
34
, G. Fontaine
28
, A. F
¨
orster
2
,
M. F
¨
ußling
35
, S. Gabici
29
, M. Gajdus
6
, Y.A. Gallant
17
, T. Garrigoux
16
, G. Giavitto
35
, B. Giebels
28
,
J.F. Glicenstein
18
, D. Gottschall
27
, A. Goyal
36
, M.-H. Grondin
37
, M. Grudzi
´
nska
19
, D. Hadasch
12
,
S. H
¨
affner
38
, J. Hahn
2
, J. Hawkes
13
, G. Heinzelmann
1
, G. Henri
30
, G. Hermann
2
, O. Hervet
15
,
A. Hillert
2
, J.A. Hinton
2,31
, W. Hofmann
2
, P. Hofverberg
2
, C. Hoischen
33
, M. Holler
28
, D. Horns
1
,
A. Ivascenko
14
, A. Jacholkowska
16
, M. Jamrozy
36
, M. Janiak
32
, F. Jankowsky
24
, I. Jung-Richardt
38
,
M.A. Kastendieck
1
, K. Katarzy
´
nski
39
, U. Katz
38
, D. Kerszberg
16
, B. Kh
´
elifi
29
, M. Kieffer
16
,
S. Klepser
35
, D. Klochkov
27
, W. Klu
´
zniak
32
, D. Kolitzus
12
, Nu. Komin
23
, K. Kosack
18
, S. Krakau
10
,
F. Krayzel
34
, P.P. Kr
¨
uger
14
, H. Laffon
37
, G. Lamanna
34
, J. Lau
13
, J. Lefaucheur
29
, V. Lefranc
18
,
A. Lemi
`
ere
29
, M. Lemoine-Goumard
37
, J.-P. Lenain
16
, T. Lohse
6
, A. Lopatin
38
, C.-C. Lu
2
, R. Lui
2
,
V. Marandon
2
, A. Marcowith
17
, C. Mariaud
28
, R. Marx
2
, G. Maurin
34
, N. Maxted
17
, M. Mayer
6
,
P.J. Meintjes
40
, U. Menzler
10
, M. Meyer
25
, A.M.W. Mitchell
2
, R. Moderski
32
, M. Mohamed
24
,
K. Mor
˚
a
25
, E. Moulin
18
, T. Murach
6
, M. de Naurois
28
, J. Niemiec
21
, L. Oakes
6
, H. Odaka
2
,
S.
¨
Ottl
12
, S. Ohm
35
, B. Opitz
1
, M. Ostrowski
36
, I. Oya
35
, M. Panter
2
, R.D. Parsons
2
, M. Paz Arribas
6
,
N.W. Pekeur
14
, G. Pelletier
30
, P.-O. Petrucci
30
, B. Peyaud
18
, S. Pita
29
, H. Poon
2
, H. Prokoph
9
,
G. P
¨
uhlhofer
27
, M. Punch
29
, A. Quirrenbach
24
, S. Raab
38
, I. Reichardt
29
, A. Reimer
12
, O. Reimer
12
,
M. Renaud
17
, R. de los Reyes
2
, F. Rieger
2,41
, C. Romoli
3
, S. Rosier-Lees
34
, G. Rowell
13
, B. Rudak
32
,
C.B. Rulten
15
, V. Sahakian
5,4
, D. Salek
42
, D.A. Sanchez
34
, A. Santangelo
27
, M. Sasaki
27
, R. Schlickeiser
10
,
F. Sch
¨
ussler
18
, A. Schulz
35
, U. Schwanke
6
, S. Schwemmer
24
, A.S. Seyffert
14
, R. Simoni
8
, H. Sol
15
,
F. Spanier
14
, G. Spengler
25
, F. Spies
1
, Ł. Stawarz
36
, R. Steenkamp
7
, C. Stegmann
33,35
, F. Stinzing
38
,
K. Stycz
35
, I. Sushch
14
, J.-P. Tavernet
16
, T. Tavernier
29
, A.M. Taylor
3
, R. Terrier
29
, M. Tluczykont
1
,
C. Trichard
34
, R. Tuffs
2
, K. Valerius
38
, J. van der Walt
14
, C. van Eldik
38
, B. van Soelen
40
, G. Vasileiadis
17
,
J. Veh
38
, C. Venter
14
, A. Viana
2
, P. Vincent
16
, J. Vink
8
, F. Voisin
13
, H.J. V
¨
olk
2
, T. Vuillaume
30
,
S.J. Wagner
24
, P. Wagner
6
, R.M. Wagner
25
, M. Weidinger
10
, Q. Weitzel
2
, R. White
31
, A. Wierzcholska
24,21
,
P. Willmann
38
, A. W
¨
ornlein
38
, D. Wouters
18
, R. Yang
2
, V. Zabalza
31
, D. Zaborov
28
, M. Zacharias
24
,
A.A. Zdziarski
32
, A. Zech
15
, F. Zefi
28
, N.
˙
Zywucka
36
1
Universit
¨
at Hamburg, Institut f
¨
ur Experimentalphysik, Luruper Chaussee 149, D 22761 Hamburg, Ger-
many
2
Max-Planck-Institut f
¨
ur Kernphysik, P.O. Box 103980, D 69029 Heidelberg, Germany
3
Dublin Institute for Advanced Studies, 31 Fitzwilliam Place, Dublin 2, Ireland
1
arXiv:1603.07730v1 [astro-ph.HE] 24 Mar 2016
4
National Academy of Sciences of the Republic of Armenia, Marshall Baghramian Avenue, 24, 0019 Yere-
van, Republic of Armenia
5
Yerevan Physics Institute, 2 Alikhanian Brothers St., 375036 Yerevan, Armenia
6
Institut f
¨
ur Physik, Humboldt-Universit
¨
at zu Berlin, Newtonstr. 15, D 12489 Berlin, Germany
7
University of Namibia, Department of Physics, Private Bag 13301, Windhoek, Namibia
8
GRAPPA, Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098
XH Amsterdam, The Netherlands
9
Department of Physics and Electrical Engineering, Linnaeus University, 351 95 V
¨
axj
¨
o, Sweden
10
Institut f
¨
ur Theoretische Physik, Lehrstuhl IV: Weltraum und Astrophysik, Ruhr-Universit
¨
at Bochum, D
44780 Bochum, Germany
11
GRAPPA, Anton Pannekoek Institute for Astronomy and Institute of High-Energy Physics, University of
Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands
12
Institut f
¨
ur Astro- und Teilchenphysik, Leopold-Franzens-Universit
¨
at Innsbruck, A-6020 Innsbruck, Aus-
tria
13
School of Chemistry & Physics, University of Adelaide, Adelaide 5005, Australia
14
Centre for Space Research, North-West University, Potchefstroom 2520, South Africa
15
LUTH, Observatoire de Paris, CNRS, Universit
´
e Paris Diderot, 5 Place Jules Janssen, 92190 Meudon,
France
16
LPNHE, Universit
´
e Pierre et Marie Curie Paris 6, Universit
´
e Denis Diderot Paris 7, CNRS/IN2P3, 4 Place
Jussieu, F-75252, Paris Cedex 5, France
17
Laboratoire Univers et Particules de Montpellier, Universit
´
e Montpellier 2, CNRS/IN2P3, CC 72, Place
Eug
`
ene Bataillon, F-34095 Montpellier Cedex 5, France
18
DSM/Irfu, CEA Saclay, F-91191 Gif-Sur-Yvette Cedex, France
19
Astronomical Observatory, The University of Warsaw, Al. Ujazdowskie 4, 00-478 Warsaw, Poland
20
Aix Marseille Universi
´
e, CNRS/IN2P3, CPPM UMR 7346, 13288 Marseille, France
21
Instytut Fizyki Ja¸drowej PAN, ul. Radzikowskiego 152, 31-342 Krak
´
ow, Poland
22
Funded by EU FP7 Marie Curie, grant agreement No. PIEF-GA-2012-332350,
23
School of Physics, University of the Witwatersrand, 1 Jan Smuts Avenue, Braamfontein, Johannesburg,
2050 South Africa
24
Landessternwarte, Universit
¨
at Heidelberg, K
¨
onigstuhl, D 69117 Heidelberg, Germany
25
Oskar Klein Centre, Department of Physics, Stockholm University, Albanova University Center, SE-
10691 Stockholm, Sweden
26
Wallenberg Academy Fellow,
27
Institut f
¨
ur Astronomie und Astrophysik, Universit
¨
at T
¨
ubingen, Sand 1, D 72076 T
¨
ubingen, Germany
28
Laboratoire Leprince-Ringuet, Ecole Polytechnique, CNRS/IN2P3, F-91128 Palaiseau, France
29
APC, AstroParticule et Cosmologie, Universit
´
e Paris Diderot, CNRS/IN2P3, CEA/Irfu, Observatoire de
Paris, Sorbonne Paris Cit
´
e, 10, rue Alice Domon et L
´
eonie Duquet, 75205 Paris Cedex 13, France
30
Univ. Grenoble Alpes, IPAG, F-38000 Grenoble, France
CNRS, IPAG, F-38000 Grenoble, France
31
Department of Physics and Astronomy, The University of Leicester, University Road, Leicester, LE1 7RH,
United Kingdom
32
Nicolaus Copernicus Astronomical Center, ul. Bartycka 18, 00-716 Warsaw, Poland
33
Institut f
¨
ur Physik und Astronomie, Universit
¨
at Potsdam, Karl-Liebknecht-Strasse 24/25, D 14476 Pots-
dam, Germany
34
Laboratoire d’Annecy-le-Vieux de Physique des Particules, Universit
´
e Savoie Mont-Blanc, CNRS/IN2P3,
2
F-74941 Annecy-le-Vieux, France
35
DESY, D-15738 Zeuthen, Germany
36
Obserwatorium Astronomiczne, Uniwersytet Jagiello
´
nski, ul. Orla 171, 30-244 Krak
´
ow, Poland
37
Universit
´
e Bordeaux, CNRS/IN2P3, Centre d’
´
Etudes Nucl
´
eaires de Bordeaux Gradignan, 33175 Gradig-
nan, France
38
Universit
¨
at Erlangen-N
¨
urnberg, Physikalisches Institut, Erwin-Rommel-Str. 1, D 91058 Erlangen, Ger-
many
39
Centre for Astronomy, Faculty of Physics, Astronomy and Informatics, Nicolaus Copernicus University,
Grudziadzka 5, 87-100 Torun, Poland
40
Department of Physics, University of the Free State, PO Box 339, Bloemfontein 9300, South Africa
41
Heisenberg Fellow (DFG), ITA Universit
¨
at Heidelberg, Germany,
42
GRAPPA, Institute of High-Energy Physics, University of Amsterdam, Science Park 904, 1098 XH Am-
sterdam, The Netherlands
Galactic cosmic rays reach energies of at least a few Peta-electronvolts (1 PeV =10
15
elec-
tron volts)
1
. This implies our Galaxy contains PeV accelerators (PeVatrons), but all proposed
models of Galactic cosmic-ray accelerators encounter non-trivial difficulties at exactly these
energies
2
. Tens of Galactic accelerators capable of accelerating particle to tens of TeV (1 TeV
=10
12
electron volts) energies were inferred from recent gamma-ray observations
3
. None
of the currently known accelerators, however, not even the handful of shell-type supernova
remnants commonly believed to supply most Galactic cosmic rays, have shown the character-
istic tracers of PeV particles: power-law spectra of gamma rays extending without a cutoff
or a spectral break to tens of TeV
4
. Here we report deep gamma-ray observations with ar-
cminute angular resolution of the Galactic Centre regions, which show the expected tracer
of the presence of PeV particles within the central 10 parsec of the Galaxy. We argue that
the supermassive black hole Sagittarius A* is linked to this PeVatron. Sagittarius A* went
through active phases in the past, as demonstrated by X-ray outbursts
5
and an outflow from
the Galactic Center
6
. Although its current rate of particle acceleration is not sufficient to
provide a substantial contribution to Galactic cosmic rays, Sagittarius A* could have plau-
sibly been more active over the last & 10
6−7
years, and therefore should be considered as a
viable alternative to supernova remnants as a source of PeV Galactic cosmic rays.
The large photon statistics accumulated over the last 10 years of observations with the High
Energy Stereoscopic System (H.E.S.S.), together with improvements in the methods of data anal-
ysis, allow for a deep study of the properties of the diffuse very-high-energy (VHE; more than 100
GeV) emission of the central molecular zone. This region surrounding the Galactic Centre contains
predominantly molecular gas and extends (in projection) out to r∼250 pc at positive galactic longi-
tudes and r∼150 pc at negative longitudes. The map of the central molecular zone as seen in VHE
γ-rays (Fig. 1) shows a strong (although not linear; see below) correlation between the brightness
distribution of VHE γ-rays and the locations of massive gas-rich complexes. This points towards a
hadronic origin of the diffuse emission
7
, where the γ-rays result from the interactions of relativis-
tic protons with the ambient gas. The second important mechanism of production of VHE γ-rays
3
is the inverse Compton scattering of electrons. However, the severe radiative losses suffered by
multi-TeV electrons in the Galactic Centre region prevent them from propagating over scales com-
parable to the size of the central molecular zone, thus disfavouring a leptonic origin of the γ-rays
(see discussion in Methods and Extended Data Figures 1 and 2).
The location and the particle injection rate history of the cosmic-ray accelerator(s), respon-
sible for the relativistic protons, determine the spatial distribution of these cosmic rays which,
together with the gas distribution, shape the morphology of the central molecular zone seen in
VHE γ-rays. Fig. 2 shows the radial profile of the E ≥ 10 TeV cosmic-rays energy density w
CR
up to r ∼ 200 pc (for a Galactic Centre distance of 8.5 kpc), determined from the γ-ray luminosity
and the amount of target gas (see Extended Data Tables 1 and 2). This high energy density in the
central molecular zone is found to be an order of magnitude larger than that of the “sea” of cosmic
rays that universally fills the Galaxy, while the energy density of low energy (GeV) cosmic rays in
this region has a level comparable to it
8
. This requires the presence of one or more accelerators of
multi-TeV particles operating in the central molecular zone.
If the accelerator injects particles at a continuous rate,
˙
Q
p
(E), the radial distribution of
cosmic rays in the central molecular zone, in the case of diffusive propagation, is described as
w
CR
(E, r, t) =
˙
Q
p
(E)
4πD(E)r
erfc(r/r
diff
)
9
, where D(E) and r
diff
are the diffusion coefficient and ra-
dius, respectively. For timescales smaller than the proton-proton interaction time in the hydrogen
gas of density n, t ≤ t
pp
' 5 × 10
4
(n/10
3
cm
−3
)
−1
yr, the diffusion radius is r
diff
≈
p
4D(E)t.
Thus, at distances r < r
diff
, the proton flux should decrease as ∼ 1/r provided that the diffusion
coefficient does not significantly vary throughout the central molecular zone. The measurements
clearly support the w
CR
(r) ∝ 1/r dependence over the entire central molecular zone region (Fig.
2) and disfavour a w
CR
(r) ∝ 1/r
2
and a w
CR
(r) ∝ constant profiles, as expected if cosmic rays
are advected in a wind, and in the case of a single burst-like event of cosmic-ray injection, respec-
tively. The 1/r profile of the cosmic-ray density up to 200 pc indicates a quasi-continuous injection
of protons into the central molecular zone from a centrally located accelerator on a timescale ∆t
exceeding the characteristic time of diffusive escape of particles from the central molecular zone,
i.e. ∆t ≥ t
diff
≈ R
2
/6D ≈ 2 × 10
3
(D/10
30
cm
2
s
−1
)
−1
yr, where D is normalised to the charac-
teristic value of multi-TeV cosmic rays in the Galactic Disk
10
. In this regime the average injection
rate of particles is found to be
˙
Q
p
(≥ 10 TeV) ≈ 4 × 10
37
(D/10
30
cm
2
s
−1
) erg/s. The diffusion
coefficient itself depends on the power spectrum of the turbulent magnetic field, which is highly
unknown in the central molecular zone region. This introduces an uncertainty in the estimates
of the injection power of relativistic protons. Yet, the diffusive nature of the propagation is con-
strained by the condition R
2
/6D R/c. For the radius of the central molecular zone region of
200 pc, this implies D 3 × 10
30
cm
2
/s, and, consequently,
˙
Q
p
1.2 × 10
38
erg/s.
The energy spectrum of the diffuse γ-ray emission (Fig. 3) has been extracted from an
annulus centred at Sagittarius (Sgr) A* (see Fig. 1). The best-fit to the data is found for a spectrum
following a power law extending with a photon index ≈2.3 to energies up to tens of TeV, without
any indication of a cutoff or a break. This is the first time that such a γ-ray spectrum, arising from
hadronic interactions, is detected in general. Since these γ-rays result from the decay of neutral
4
pions produced by pp interactions, the derivation of such hard power-law spectrum implies that
the spectrum of the parent protons should extend to energies close to 1 PeV. The best fit of a γ-ray
spectrum from neutral pion decay to the H.E.S.S. data is found for a proton spectrum following
a pure power-law with index ≈2.4. We note that pp interactions of 1 PeV protons could also be
studied by the observation of emitted neutrinos or the X-rays from the synchrotron emission of
secondary electrons and positrons (see Methods and Extended Data Figures 3 and 4). However,
the measured γ-ray flux puts the expected fluxes of neutrinos and X-rays below or at best close to
the sensitivities of the current instruments. Assuming a cutoff in the parent proton spectrum, the
corresponding secondary γ-ray spectrum deviates from the H.E.S.S. data at 68%, 90% and 95%
confidence levels for cutoffs at 2.9 PeV, 0.6 PeV and 0.4 PeV, respectively. This is the first robust
detection of a VHE cosmic hadronic accelerator which operates as a PeVatron.
Remarkably, the Galactic Centre PeVatron appears to be located in the same region as the
central γ-ray source HESS J1745-290
11–14
. Unfortunately, the current data cannot provide an answer
as to whether there is an intrinsic link between these two objects. The point-like source HESS
J1745-290 itself remains unidentified. Besides Sgr A*
15
, other potential counterparts are the pulsar
wind nebula G 359.95-0.04
16, 17
, and a spike of annihilating dark matter
18
. Moreover, it has also
been suggested that this source might have a diffuse origin, peaking towards the direction of the
Galactic Centreg because of the higher concentration of both gas and relativistic particles
15
. In
fact, this interpretation would imply an extension of the spectrum of the central source to energies
beyond 10 TeV, which however is at odds with the detection of a clear cutoff in the spectrum of
HESS J1745-290 at about 10 TeV
19, 20
(Fig. 3). Yet, the attractive idea of explaining the entire γ-ray
emission from the Galactic Centre by run-away protons from the same centrally located accelerator
can be still compatible with the cutoff in the spectrum of the central source. For example, the cutoff
could be due to the absorption of γ-rays caused by interactions with the ambient infrared radiation
field. It should be noted that although the question on the link between the central γ-ray source
and the proton PeVatron is an interesting issue in its own right, it, however, does not have a direct
impact on the main conclusions of this work.
The integration of the cosmic-ray radial distribution (Fig. 2) yields the total energy of E≥10
TeV protons confined in the central molecular zone: W
CR
≈ 1.0 × 10
49
erg. A single Supernova
Remnant (SNR) would suffice to provide this rather modest energy in cosmic rays. A possible
candidate could be Sgr A East. Although this object has already been excluded as a counterpart
of HESS J1745-290
21
, the multi-TeV protons accelerated by this object and then injected into the
central molecular zone could contribute to the diffuse γ-ray component. Another potential site
for acceleration of protons in the Galactic Centre are the compact stellar clusters
22
. Formally,
the mechanical power in these clusters in the form of stellar winds, which can provide adequate
conditions for particle acceleration, is sufficient to explain the required total energy of cosmic rays
in the central molecular zone. However, the acceleration of protons to PeV energies requires bulk
motions in excess of 10,000 km/s which in the stellar clusters could only exist because of very
young supernova shocks
23
. Thus, the operation of PeVatrons in stellar clusters is reduced to the
presence of supernovae shocks. On the other hand, since the acceleration of PeV particles by
shocks, either in the individual SNRs or in the stellar clusters, cannot last significantly longer than
5
